Commit
•
204b65f
1
Parent(s):
6018af8
Upload 10 files
Browse filesupload new checkpoint
- adapter_config.json +5 -5
- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- tokenizer.json +64 -5
- trainer_state.json +3504 -4
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -20,13 +20,13 @@
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": "unsloth",
|
| 22 |
"target_modules": [
|
| 23 |
-
"
|
| 24 |
-
"q_proj",
|
| 25 |
"o_proj",
|
| 26 |
-
"
|
| 27 |
-
"gate_proj",
|
| 28 |
"up_proj",
|
| 29 |
-
"
|
|
|
|
|
|
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
| 32 |
"use_dora": false,
|
|
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": "unsloth",
|
| 22 |
"target_modules": [
|
| 23 |
+
"k_proj",
|
|
|
|
| 24 |
"o_proj",
|
| 25 |
+
"v_proj",
|
|
|
|
| 26 |
"up_proj",
|
| 27 |
+
"gate_proj",
|
| 28 |
+
"q_proj",
|
| 29 |
+
"down_proj"
|
| 30 |
],
|
| 31 |
"task_type": "CAUSAL_LM",
|
| 32 |
"use_dora": false,
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167832240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6931e308d6061f03d2e2e7100637d59626c76118a903132beaa3ae6edbeacaae
|
| 3 |
size 167832240
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 85736914
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5eb205e9a3d8cd1af73c6e0c3e12eb5b93ffb579b4f097e0a6f1109c78e360b9
|
| 3 |
size 85736914
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ae282cbf9290d2263341594687ec3f780027ee4e703cfdf970362e3bc87775ec
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08de451528ec8f4b0e224d4376c47d818e1cf4afb1e94fd23999f7fa09c11203
|
| 3 |
size 1064
|
tokenizer.json
CHANGED
|
@@ -2329,10 +2329,69 @@
|
|
| 2329 |
]
|
| 2330 |
},
|
| 2331 |
"post_processor": {
|
| 2332 |
-
"type": "
|
| 2333 |
-
"
|
| 2334 |
-
|
| 2335 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2336 |
},
|
| 2337 |
"decoder": {
|
| 2338 |
"type": "ByteLevel",
|
|
@@ -2348,7 +2407,7 @@
|
|
| 2348 |
"end_of_word_suffix": null,
|
| 2349 |
"fuse_unk": false,
|
| 2350 |
"byte_fallback": false,
|
| 2351 |
-
"ignore_merges":
|
| 2352 |
"vocab": {
|
| 2353 |
"!": 0,
|
| 2354 |
"\"": 1,
|
|
|
|
| 2329 |
]
|
| 2330 |
},
|
| 2331 |
"post_processor": {
|
| 2332 |
+
"type": "Sequence",
|
| 2333 |
+
"processors": [
|
| 2334 |
+
{
|
| 2335 |
+
"type": "ByteLevel",
|
| 2336 |
+
"add_prefix_space": true,
|
| 2337 |
+
"trim_offsets": false,
|
| 2338 |
+
"use_regex": true
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"type": "TemplateProcessing",
|
| 2342 |
+
"single": [
|
| 2343 |
+
{
|
| 2344 |
+
"SpecialToken": {
|
| 2345 |
+
"id": "<|begin_of_text|>",
|
| 2346 |
+
"type_id": 0
|
| 2347 |
+
}
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"Sequence": {
|
| 2351 |
+
"id": "A",
|
| 2352 |
+
"type_id": 0
|
| 2353 |
+
}
|
| 2354 |
+
}
|
| 2355 |
+
],
|
| 2356 |
+
"pair": [
|
| 2357 |
+
{
|
| 2358 |
+
"SpecialToken": {
|
| 2359 |
+
"id": "<|begin_of_text|>",
|
| 2360 |
+
"type_id": 0
|
| 2361 |
+
}
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"Sequence": {
|
| 2365 |
+
"id": "A",
|
| 2366 |
+
"type_id": 0
|
| 2367 |
+
}
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"SpecialToken": {
|
| 2371 |
+
"id": "<|begin_of_text|>",
|
| 2372 |
+
"type_id": 1
|
| 2373 |
+
}
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"Sequence": {
|
| 2377 |
+
"id": "B",
|
| 2378 |
+
"type_id": 1
|
| 2379 |
+
}
|
| 2380 |
+
}
|
| 2381 |
+
],
|
| 2382 |
+
"special_tokens": {
|
| 2383 |
+
"<|begin_of_text|>": {
|
| 2384 |
+
"id": "<|begin_of_text|>",
|
| 2385 |
+
"ids": [
|
| 2386 |
+
128000
|
| 2387 |
+
],
|
| 2388 |
+
"tokens": [
|
| 2389 |
+
"<|begin_of_text|>"
|
| 2390 |
+
]
|
| 2391 |
+
}
|
| 2392 |
+
}
|
| 2393 |
+
}
|
| 2394 |
+
]
|
| 2395 |
},
|
| 2396 |
"decoder": {
|
| 2397 |
"type": "ByteLevel",
|
|
|
|
| 2407 |
"end_of_word_suffix": null,
|
| 2408 |
"fuse_unk": false,
|
| 2409 |
"byte_fallback": false,
|
| 2410 |
+
"ignore_merges": true,
|
| 2411 |
"vocab": {
|
| 2412 |
"!": 0,
|
| 2413 |
"\"": 1,
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 1.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7007,14 +7007,3514 @@
|
|
| 7007 |
"learning_rate": 8.298999411418483e-05,
|
| 7008 |
"loss": 0.6655,
|
| 7009 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7010 |
}
|
| 7011 |
],
|
| 7012 |
"logging_steps": 1,
|
| 7013 |
-
"max_steps":
|
| 7014 |
"num_input_tokens_seen": 0,
|
| 7015 |
"num_train_epochs": 2,
|
| 7016 |
"save_steps": 100,
|
| 7017 |
-
"total_flos":
|
| 7018 |
"train_batch_size": 4,
|
| 7019 |
"trial_name": null,
|
| 7020 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.7582771755054205,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7007 |
"learning_rate": 8.298999411418483e-05,
|
| 7008 |
"loss": 0.6655,
|
| 7009 |
"step": 1000
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 1.1748826291079812,
|
| 7013 |
+
"grad_norm": 0.37893033027648926,
|
| 7014 |
+
"learning_rate": 8.287227781047675e-05,
|
| 7015 |
+
"loss": 0.7422,
|
| 7016 |
+
"step": 1001
|
| 7017 |
+
},
|
| 7018 |
+
{
|
| 7019 |
+
"epoch": 1.176056338028169,
|
| 7020 |
+
"grad_norm": 0.3806493580341339,
|
| 7021 |
+
"learning_rate": 8.27545615067687e-05,
|
| 7022 |
+
"loss": 0.7147,
|
| 7023 |
+
"step": 1002
|
| 7024 |
+
},
|
| 7025 |
+
{
|
| 7026 |
+
"epoch": 1.1772300469483568,
|
| 7027 |
+
"grad_norm": 0.36736544966697693,
|
| 7028 |
+
"learning_rate": 8.263684520306063e-05,
|
| 7029 |
+
"loss": 0.6813,
|
| 7030 |
+
"step": 1003
|
| 7031 |
+
},
|
| 7032 |
+
{
|
| 7033 |
+
"epoch": 1.1784037558685445,
|
| 7034 |
+
"grad_norm": 0.3947063088417053,
|
| 7035 |
+
"learning_rate": 8.251912889935256e-05,
|
| 7036 |
+
"loss": 0.6863,
|
| 7037 |
+
"step": 1004
|
| 7038 |
+
},
|
| 7039 |
+
{
|
| 7040 |
+
"epoch": 1.1795774647887325,
|
| 7041 |
+
"grad_norm": 0.3787820339202881,
|
| 7042 |
+
"learning_rate": 8.24014125956445e-05,
|
| 7043 |
+
"loss": 0.6929,
|
| 7044 |
+
"step": 1005
|
| 7045 |
+
},
|
| 7046 |
+
{
|
| 7047 |
+
"epoch": 1.1807511737089202,
|
| 7048 |
+
"grad_norm": 0.3705902099609375,
|
| 7049 |
+
"learning_rate": 8.228369629193645e-05,
|
| 7050 |
+
"loss": 0.6536,
|
| 7051 |
+
"step": 1006
|
| 7052 |
+
},
|
| 7053 |
+
{
|
| 7054 |
+
"epoch": 1.181924882629108,
|
| 7055 |
+
"grad_norm": 0.38193127512931824,
|
| 7056 |
+
"learning_rate": 8.216597998822837e-05,
|
| 7057 |
+
"loss": 0.6822,
|
| 7058 |
+
"step": 1007
|
| 7059 |
+
},
|
| 7060 |
+
{
|
| 7061 |
+
"epoch": 1.1830985915492958,
|
| 7062 |
+
"grad_norm": 0.3939112424850464,
|
| 7063 |
+
"learning_rate": 8.204826368452032e-05,
|
| 7064 |
+
"loss": 0.7211,
|
| 7065 |
+
"step": 1008
|
| 7066 |
+
},
|
| 7067 |
+
{
|
| 7068 |
+
"epoch": 1.1842723004694835,
|
| 7069 |
+
"grad_norm": 0.3711683452129364,
|
| 7070 |
+
"learning_rate": 8.193054738081225e-05,
|
| 7071 |
+
"loss": 0.6839,
|
| 7072 |
+
"step": 1009
|
| 7073 |
+
},
|
| 7074 |
+
{
|
| 7075 |
+
"epoch": 1.1854460093896715,
|
| 7076 |
+
"grad_norm": 0.3889683187007904,
|
| 7077 |
+
"learning_rate": 8.181283107710418e-05,
|
| 7078 |
+
"loss": 0.7151,
|
| 7079 |
+
"step": 1010
|
| 7080 |
+
},
|
| 7081 |
+
{
|
| 7082 |
+
"epoch": 1.1866197183098592,
|
| 7083 |
+
"grad_norm": 0.3716399073600769,
|
| 7084 |
+
"learning_rate": 8.169511477339612e-05,
|
| 7085 |
+
"loss": 0.6706,
|
| 7086 |
+
"step": 1011
|
| 7087 |
+
},
|
| 7088 |
+
{
|
| 7089 |
+
"epoch": 1.187793427230047,
|
| 7090 |
+
"grad_norm": 0.3653043806552887,
|
| 7091 |
+
"learning_rate": 8.157739846968805e-05,
|
| 7092 |
+
"loss": 0.691,
|
| 7093 |
+
"step": 1012
|
| 7094 |
+
},
|
| 7095 |
+
{
|
| 7096 |
+
"epoch": 1.1889671361502347,
|
| 7097 |
+
"grad_norm": 0.37435877323150635,
|
| 7098 |
+
"learning_rate": 8.145968216597999e-05,
|
| 7099 |
+
"loss": 0.6431,
|
| 7100 |
+
"step": 1013
|
| 7101 |
+
},
|
| 7102 |
+
{
|
| 7103 |
+
"epoch": 1.1901408450704225,
|
| 7104 |
+
"grad_norm": 0.36023011803627014,
|
| 7105 |
+
"learning_rate": 8.134196586227192e-05,
|
| 7106 |
+
"loss": 0.6397,
|
| 7107 |
+
"step": 1014
|
| 7108 |
+
},
|
| 7109 |
+
{
|
| 7110 |
+
"epoch": 1.1913145539906103,
|
| 7111 |
+
"grad_norm": 0.3797522485256195,
|
| 7112 |
+
"learning_rate": 8.122424955856387e-05,
|
| 7113 |
+
"loss": 0.6615,
|
| 7114 |
+
"step": 1015
|
| 7115 |
+
},
|
| 7116 |
+
{
|
| 7117 |
+
"epoch": 1.192488262910798,
|
| 7118 |
+
"grad_norm": 0.3832303285598755,
|
| 7119 |
+
"learning_rate": 8.11065332548558e-05,
|
| 7120 |
+
"loss": 0.6685,
|
| 7121 |
+
"step": 1016
|
| 7122 |
+
},
|
| 7123 |
+
{
|
| 7124 |
+
"epoch": 1.193661971830986,
|
| 7125 |
+
"grad_norm": 0.37008216977119446,
|
| 7126 |
+
"learning_rate": 8.098881695114774e-05,
|
| 7127 |
+
"loss": 0.7051,
|
| 7128 |
+
"step": 1017
|
| 7129 |
+
},
|
| 7130 |
+
{
|
| 7131 |
+
"epoch": 1.1948356807511737,
|
| 7132 |
+
"grad_norm": 0.3758820593357086,
|
| 7133 |
+
"learning_rate": 8.087110064743967e-05,
|
| 7134 |
+
"loss": 0.6753,
|
| 7135 |
+
"step": 1018
|
| 7136 |
+
},
|
| 7137 |
+
{
|
| 7138 |
+
"epoch": 1.1960093896713615,
|
| 7139 |
+
"grad_norm": 0.3715493977069855,
|
| 7140 |
+
"learning_rate": 8.075338434373161e-05,
|
| 7141 |
+
"loss": 0.6863,
|
| 7142 |
+
"step": 1019
|
| 7143 |
+
},
|
| 7144 |
+
{
|
| 7145 |
+
"epoch": 1.1971830985915493,
|
| 7146 |
+
"grad_norm": 0.36335518956184387,
|
| 7147 |
+
"learning_rate": 8.063566804002354e-05,
|
| 7148 |
+
"loss": 0.6865,
|
| 7149 |
+
"step": 1020
|
| 7150 |
+
},
|
| 7151 |
+
{
|
| 7152 |
+
"epoch": 1.198356807511737,
|
| 7153 |
+
"grad_norm": 0.36738407611846924,
|
| 7154 |
+
"learning_rate": 8.051795173631549e-05,
|
| 7155 |
+
"loss": 0.6729,
|
| 7156 |
+
"step": 1021
|
| 7157 |
+
},
|
| 7158 |
+
{
|
| 7159 |
+
"epoch": 1.199530516431925,
|
| 7160 |
+
"grad_norm": 0.37416642904281616,
|
| 7161 |
+
"learning_rate": 8.040023543260741e-05,
|
| 7162 |
+
"loss": 0.6904,
|
| 7163 |
+
"step": 1022
|
| 7164 |
+
},
|
| 7165 |
+
{
|
| 7166 |
+
"epoch": 1.2007042253521127,
|
| 7167 |
+
"grad_norm": 0.37728917598724365,
|
| 7168 |
+
"learning_rate": 8.028251912889936e-05,
|
| 7169 |
+
"loss": 0.6613,
|
| 7170 |
+
"step": 1023
|
| 7171 |
+
},
|
| 7172 |
+
{
|
| 7173 |
+
"epoch": 1.2018779342723005,
|
| 7174 |
+
"grad_norm": 0.3821254074573517,
|
| 7175 |
+
"learning_rate": 8.016480282519128e-05,
|
| 7176 |
+
"loss": 0.6567,
|
| 7177 |
+
"step": 1024
|
| 7178 |
+
},
|
| 7179 |
+
{
|
| 7180 |
+
"epoch": 1.2030516431924883,
|
| 7181 |
+
"grad_norm": 0.37766242027282715,
|
| 7182 |
+
"learning_rate": 8.004708652148323e-05,
|
| 7183 |
+
"loss": 0.6648,
|
| 7184 |
+
"step": 1025
|
| 7185 |
+
},
|
| 7186 |
+
{
|
| 7187 |
+
"epoch": 1.204225352112676,
|
| 7188 |
+
"grad_norm": 0.3994279205799103,
|
| 7189 |
+
"learning_rate": 7.992937021777516e-05,
|
| 7190 |
+
"loss": 0.7085,
|
| 7191 |
+
"step": 1026
|
| 7192 |
+
},
|
| 7193 |
+
{
|
| 7194 |
+
"epoch": 1.2053990610328638,
|
| 7195 |
+
"grad_norm": 0.3634461760520935,
|
| 7196 |
+
"learning_rate": 7.98116539140671e-05,
|
| 7197 |
+
"loss": 0.69,
|
| 7198 |
+
"step": 1027
|
| 7199 |
+
},
|
| 7200 |
+
{
|
| 7201 |
+
"epoch": 1.2065727699530517,
|
| 7202 |
+
"grad_norm": 0.36296284198760986,
|
| 7203 |
+
"learning_rate": 7.969393761035903e-05,
|
| 7204 |
+
"loss": 0.6394,
|
| 7205 |
+
"step": 1028
|
| 7206 |
+
},
|
| 7207 |
+
{
|
| 7208 |
+
"epoch": 1.2077464788732395,
|
| 7209 |
+
"grad_norm": 0.38547366857528687,
|
| 7210 |
+
"learning_rate": 7.957622130665098e-05,
|
| 7211 |
+
"loss": 0.6602,
|
| 7212 |
+
"step": 1029
|
| 7213 |
+
},
|
| 7214 |
+
{
|
| 7215 |
+
"epoch": 1.2089201877934272,
|
| 7216 |
+
"grad_norm": 0.38083600997924805,
|
| 7217 |
+
"learning_rate": 7.94585050029429e-05,
|
| 7218 |
+
"loss": 0.6559,
|
| 7219 |
+
"step": 1030
|
| 7220 |
+
},
|
| 7221 |
+
{
|
| 7222 |
+
"epoch": 1.210093896713615,
|
| 7223 |
+
"grad_norm": 0.377316415309906,
|
| 7224 |
+
"learning_rate": 7.934078869923485e-05,
|
| 7225 |
+
"loss": 0.7096,
|
| 7226 |
+
"step": 1031
|
| 7227 |
+
},
|
| 7228 |
+
{
|
| 7229 |
+
"epoch": 1.2112676056338028,
|
| 7230 |
+
"grad_norm": 0.37379539012908936,
|
| 7231 |
+
"learning_rate": 7.922307239552679e-05,
|
| 7232 |
+
"loss": 0.7321,
|
| 7233 |
+
"step": 1032
|
| 7234 |
+
},
|
| 7235 |
+
{
|
| 7236 |
+
"epoch": 1.2124413145539905,
|
| 7237 |
+
"grad_norm": 0.3899437487125397,
|
| 7238 |
+
"learning_rate": 7.910535609181872e-05,
|
| 7239 |
+
"loss": 0.6307,
|
| 7240 |
+
"step": 1033
|
| 7241 |
+
},
|
| 7242 |
+
{
|
| 7243 |
+
"epoch": 1.2136150234741785,
|
| 7244 |
+
"grad_norm": 0.3994787931442261,
|
| 7245 |
+
"learning_rate": 7.898763978811065e-05,
|
| 7246 |
+
"loss": 0.7189,
|
| 7247 |
+
"step": 1034
|
| 7248 |
+
},
|
| 7249 |
+
{
|
| 7250 |
+
"epoch": 1.2147887323943662,
|
| 7251 |
+
"grad_norm": 0.3705124855041504,
|
| 7252 |
+
"learning_rate": 7.88699234844026e-05,
|
| 7253 |
+
"loss": 0.6683,
|
| 7254 |
+
"step": 1035
|
| 7255 |
+
},
|
| 7256 |
+
{
|
| 7257 |
+
"epoch": 1.215962441314554,
|
| 7258 |
+
"grad_norm": 0.39496761560440063,
|
| 7259 |
+
"learning_rate": 7.875220718069452e-05,
|
| 7260 |
+
"loss": 0.6961,
|
| 7261 |
+
"step": 1036
|
| 7262 |
+
},
|
| 7263 |
+
{
|
| 7264 |
+
"epoch": 1.2171361502347418,
|
| 7265 |
+
"grad_norm": 0.38047799468040466,
|
| 7266 |
+
"learning_rate": 7.863449087698647e-05,
|
| 7267 |
+
"loss": 0.6877,
|
| 7268 |
+
"step": 1037
|
| 7269 |
+
},
|
| 7270 |
+
{
|
| 7271 |
+
"epoch": 1.2183098591549295,
|
| 7272 |
+
"grad_norm": 0.3772275745868683,
|
| 7273 |
+
"learning_rate": 7.85167745732784e-05,
|
| 7274 |
+
"loss": 0.6523,
|
| 7275 |
+
"step": 1038
|
| 7276 |
+
},
|
| 7277 |
+
{
|
| 7278 |
+
"epoch": 1.2194835680751175,
|
| 7279 |
+
"grad_norm": 0.37646278738975525,
|
| 7280 |
+
"learning_rate": 7.839905826957034e-05,
|
| 7281 |
+
"loss": 0.686,
|
| 7282 |
+
"step": 1039
|
| 7283 |
+
},
|
| 7284 |
+
{
|
| 7285 |
+
"epoch": 1.2206572769953052,
|
| 7286 |
+
"grad_norm": 0.3657703101634979,
|
| 7287 |
+
"learning_rate": 7.828134196586228e-05,
|
| 7288 |
+
"loss": 0.6904,
|
| 7289 |
+
"step": 1040
|
| 7290 |
+
},
|
| 7291 |
+
{
|
| 7292 |
+
"epoch": 1.221830985915493,
|
| 7293 |
+
"grad_norm": 0.36845406889915466,
|
| 7294 |
+
"learning_rate": 7.816362566215421e-05,
|
| 7295 |
+
"loss": 0.6795,
|
| 7296 |
+
"step": 1041
|
| 7297 |
+
},
|
| 7298 |
+
{
|
| 7299 |
+
"epoch": 1.2230046948356808,
|
| 7300 |
+
"grad_norm": 0.37235668301582336,
|
| 7301 |
+
"learning_rate": 7.804590935844614e-05,
|
| 7302 |
+
"loss": 0.6406,
|
| 7303 |
+
"step": 1042
|
| 7304 |
+
},
|
| 7305 |
+
{
|
| 7306 |
+
"epoch": 1.2241784037558685,
|
| 7307 |
+
"grad_norm": 0.3837567865848541,
|
| 7308 |
+
"learning_rate": 7.792819305473808e-05,
|
| 7309 |
+
"loss": 0.7387,
|
| 7310 |
+
"step": 1043
|
| 7311 |
+
},
|
| 7312 |
+
{
|
| 7313 |
+
"epoch": 1.2253521126760563,
|
| 7314 |
+
"grad_norm": 0.3731269836425781,
|
| 7315 |
+
"learning_rate": 7.781047675103003e-05,
|
| 7316 |
+
"loss": 0.6767,
|
| 7317 |
+
"step": 1044
|
| 7318 |
+
},
|
| 7319 |
+
{
|
| 7320 |
+
"epoch": 1.226525821596244,
|
| 7321 |
+
"grad_norm": 0.3709997534751892,
|
| 7322 |
+
"learning_rate": 7.769276044732196e-05,
|
| 7323 |
+
"loss": 0.6843,
|
| 7324 |
+
"step": 1045
|
| 7325 |
+
},
|
| 7326 |
+
{
|
| 7327 |
+
"epoch": 1.227699530516432,
|
| 7328 |
+
"grad_norm": 0.36420220136642456,
|
| 7329 |
+
"learning_rate": 7.75750441436139e-05,
|
| 7330 |
+
"loss": 0.6732,
|
| 7331 |
+
"step": 1046
|
| 7332 |
+
},
|
| 7333 |
+
{
|
| 7334 |
+
"epoch": 1.2288732394366197,
|
| 7335 |
+
"grad_norm": 0.35353943705558777,
|
| 7336 |
+
"learning_rate": 7.745732783990583e-05,
|
| 7337 |
+
"loss": 0.6955,
|
| 7338 |
+
"step": 1047
|
| 7339 |
+
},
|
| 7340 |
+
{
|
| 7341 |
+
"epoch": 1.2300469483568075,
|
| 7342 |
+
"grad_norm": 0.3663172721862793,
|
| 7343 |
+
"learning_rate": 7.733961153619777e-05,
|
| 7344 |
+
"loss": 0.6509,
|
| 7345 |
+
"step": 1048
|
| 7346 |
+
},
|
| 7347 |
+
{
|
| 7348 |
+
"epoch": 1.2312206572769953,
|
| 7349 |
+
"grad_norm": 0.3808366060256958,
|
| 7350 |
+
"learning_rate": 7.72218952324897e-05,
|
| 7351 |
+
"loss": 0.6736,
|
| 7352 |
+
"step": 1049
|
| 7353 |
+
},
|
| 7354 |
+
{
|
| 7355 |
+
"epoch": 1.232394366197183,
|
| 7356 |
+
"grad_norm": 0.3684788942337036,
|
| 7357 |
+
"learning_rate": 7.710417892878165e-05,
|
| 7358 |
+
"loss": 0.7294,
|
| 7359 |
+
"step": 1050
|
| 7360 |
+
},
|
| 7361 |
+
{
|
| 7362 |
+
"epoch": 1.233568075117371,
|
| 7363 |
+
"grad_norm": 0.3792336881160736,
|
| 7364 |
+
"learning_rate": 7.698646262507357e-05,
|
| 7365 |
+
"loss": 0.6864,
|
| 7366 |
+
"step": 1051
|
| 7367 |
+
},
|
| 7368 |
+
{
|
| 7369 |
+
"epoch": 1.2347417840375587,
|
| 7370 |
+
"grad_norm": 0.3681674897670746,
|
| 7371 |
+
"learning_rate": 7.686874632136552e-05,
|
| 7372 |
+
"loss": 0.6406,
|
| 7373 |
+
"step": 1052
|
| 7374 |
+
},
|
| 7375 |
+
{
|
| 7376 |
+
"epoch": 1.2359154929577465,
|
| 7377 |
+
"grad_norm": 0.3578929603099823,
|
| 7378 |
+
"learning_rate": 7.675103001765744e-05,
|
| 7379 |
+
"loss": 0.6799,
|
| 7380 |
+
"step": 1053
|
| 7381 |
+
},
|
| 7382 |
+
{
|
| 7383 |
+
"epoch": 1.2370892018779343,
|
| 7384 |
+
"grad_norm": 0.3885924816131592,
|
| 7385 |
+
"learning_rate": 7.663331371394939e-05,
|
| 7386 |
+
"loss": 0.6888,
|
| 7387 |
+
"step": 1054
|
| 7388 |
+
},
|
| 7389 |
+
{
|
| 7390 |
+
"epoch": 1.238262910798122,
|
| 7391 |
+
"grad_norm": 0.3622206747531891,
|
| 7392 |
+
"learning_rate": 7.651559741024132e-05,
|
| 7393 |
+
"loss": 0.6591,
|
| 7394 |
+
"step": 1055
|
| 7395 |
+
},
|
| 7396 |
+
{
|
| 7397 |
+
"epoch": 1.2394366197183098,
|
| 7398 |
+
"grad_norm": 0.36698073148727417,
|
| 7399 |
+
"learning_rate": 7.639788110653326e-05,
|
| 7400 |
+
"loss": 0.6703,
|
| 7401 |
+
"step": 1056
|
| 7402 |
+
},
|
| 7403 |
+
{
|
| 7404 |
+
"epoch": 1.2406103286384977,
|
| 7405 |
+
"grad_norm": 0.3823949992656708,
|
| 7406 |
+
"learning_rate": 7.628016480282519e-05,
|
| 7407 |
+
"loss": 0.6354,
|
| 7408 |
+
"step": 1057
|
| 7409 |
+
},
|
| 7410 |
+
{
|
| 7411 |
+
"epoch": 1.2417840375586855,
|
| 7412 |
+
"grad_norm": 0.3905463218688965,
|
| 7413 |
+
"learning_rate": 7.616244849911714e-05,
|
| 7414 |
+
"loss": 0.6875,
|
| 7415 |
+
"step": 1058
|
| 7416 |
+
},
|
| 7417 |
+
{
|
| 7418 |
+
"epoch": 1.2429577464788732,
|
| 7419 |
+
"grad_norm": 0.3658163845539093,
|
| 7420 |
+
"learning_rate": 7.604473219540906e-05,
|
| 7421 |
+
"loss": 0.6918,
|
| 7422 |
+
"step": 1059
|
| 7423 |
+
},
|
| 7424 |
+
{
|
| 7425 |
+
"epoch": 1.244131455399061,
|
| 7426 |
+
"grad_norm": 0.3887515068054199,
|
| 7427 |
+
"learning_rate": 7.592701589170101e-05,
|
| 7428 |
+
"loss": 0.6636,
|
| 7429 |
+
"step": 1060
|
| 7430 |
+
},
|
| 7431 |
+
{
|
| 7432 |
+
"epoch": 1.2453051643192488,
|
| 7433 |
+
"grad_norm": 0.38240671157836914,
|
| 7434 |
+
"learning_rate": 7.580929958799294e-05,
|
| 7435 |
+
"loss": 0.7151,
|
| 7436 |
+
"step": 1061
|
| 7437 |
+
},
|
| 7438 |
+
{
|
| 7439 |
+
"epoch": 1.2464788732394365,
|
| 7440 |
+
"grad_norm": 0.3772590458393097,
|
| 7441 |
+
"learning_rate": 7.569158328428488e-05,
|
| 7442 |
+
"loss": 0.646,
|
| 7443 |
+
"step": 1062
|
| 7444 |
+
},
|
| 7445 |
+
{
|
| 7446 |
+
"epoch": 1.2476525821596245,
|
| 7447 |
+
"grad_norm": 0.4079722464084625,
|
| 7448 |
+
"learning_rate": 7.557386698057681e-05,
|
| 7449 |
+
"loss": 0.6599,
|
| 7450 |
+
"step": 1063
|
| 7451 |
+
},
|
| 7452 |
+
{
|
| 7453 |
+
"epoch": 1.2488262910798122,
|
| 7454 |
+
"grad_norm": 0.36246946454048157,
|
| 7455 |
+
"learning_rate": 7.545615067686876e-05,
|
| 7456 |
+
"loss": 0.6512,
|
| 7457 |
+
"step": 1064
|
| 7458 |
+
},
|
| 7459 |
+
{
|
| 7460 |
+
"epoch": 1.25,
|
| 7461 |
+
"grad_norm": 0.4114508330821991,
|
| 7462 |
+
"learning_rate": 7.533843437316068e-05,
|
| 7463 |
+
"loss": 0.6756,
|
| 7464 |
+
"step": 1065
|
| 7465 |
+
},
|
| 7466 |
+
{
|
| 7467 |
+
"epoch": 1.2511737089201878,
|
| 7468 |
+
"grad_norm": 0.40775856375694275,
|
| 7469 |
+
"learning_rate": 7.522071806945263e-05,
|
| 7470 |
+
"loss": 0.7285,
|
| 7471 |
+
"step": 1066
|
| 7472 |
+
},
|
| 7473 |
+
{
|
| 7474 |
+
"epoch": 1.2523474178403755,
|
| 7475 |
+
"grad_norm": 0.3941507339477539,
|
| 7476 |
+
"learning_rate": 7.510300176574456e-05,
|
| 7477 |
+
"loss": 0.6548,
|
| 7478 |
+
"step": 1067
|
| 7479 |
+
},
|
| 7480 |
+
{
|
| 7481 |
+
"epoch": 1.2535211267605635,
|
| 7482 |
+
"grad_norm": 0.3808498978614807,
|
| 7483 |
+
"learning_rate": 7.49852854620365e-05,
|
| 7484 |
+
"loss": 0.6922,
|
| 7485 |
+
"step": 1068
|
| 7486 |
+
},
|
| 7487 |
+
{
|
| 7488 |
+
"epoch": 1.2546948356807512,
|
| 7489 |
+
"grad_norm": 0.39144501090049744,
|
| 7490 |
+
"learning_rate": 7.486756915832843e-05,
|
| 7491 |
+
"loss": 0.779,
|
| 7492 |
+
"step": 1069
|
| 7493 |
+
},
|
| 7494 |
+
{
|
| 7495 |
+
"epoch": 1.255868544600939,
|
| 7496 |
+
"grad_norm": 0.3885059356689453,
|
| 7497 |
+
"learning_rate": 7.474985285462037e-05,
|
| 7498 |
+
"loss": 0.7023,
|
| 7499 |
+
"step": 1070
|
| 7500 |
+
},
|
| 7501 |
+
{
|
| 7502 |
+
"epoch": 1.2570422535211268,
|
| 7503 |
+
"grad_norm": 0.37694790959358215,
|
| 7504 |
+
"learning_rate": 7.46321365509123e-05,
|
| 7505 |
+
"loss": 0.6306,
|
| 7506 |
+
"step": 1071
|
| 7507 |
+
},
|
| 7508 |
+
{
|
| 7509 |
+
"epoch": 1.2582159624413145,
|
| 7510 |
+
"grad_norm": 0.3846953213214874,
|
| 7511 |
+
"learning_rate": 7.451442024720424e-05,
|
| 7512 |
+
"loss": 0.6719,
|
| 7513 |
+
"step": 1072
|
| 7514 |
+
},
|
| 7515 |
+
{
|
| 7516 |
+
"epoch": 1.2593896713615025,
|
| 7517 |
+
"grad_norm": 0.3871643841266632,
|
| 7518 |
+
"learning_rate": 7.439670394349618e-05,
|
| 7519 |
+
"loss": 0.7057,
|
| 7520 |
+
"step": 1073
|
| 7521 |
+
},
|
| 7522 |
+
{
|
| 7523 |
+
"epoch": 1.26056338028169,
|
| 7524 |
+
"grad_norm": 0.37070584297180176,
|
| 7525 |
+
"learning_rate": 7.427898763978812e-05,
|
| 7526 |
+
"loss": 0.706,
|
| 7527 |
+
"step": 1074
|
| 7528 |
+
},
|
| 7529 |
+
{
|
| 7530 |
+
"epoch": 1.261737089201878,
|
| 7531 |
+
"grad_norm": 0.3774947226047516,
|
| 7532 |
+
"learning_rate": 7.416127133608005e-05,
|
| 7533 |
+
"loss": 0.6918,
|
| 7534 |
+
"step": 1075
|
| 7535 |
+
},
|
| 7536 |
+
{
|
| 7537 |
+
"epoch": 1.2629107981220657,
|
| 7538 |
+
"grad_norm": 0.36732053756713867,
|
| 7539 |
+
"learning_rate": 7.404355503237199e-05,
|
| 7540 |
+
"loss": 0.6639,
|
| 7541 |
+
"step": 1076
|
| 7542 |
+
},
|
| 7543 |
+
{
|
| 7544 |
+
"epoch": 1.2640845070422535,
|
| 7545 |
+
"grad_norm": 0.36723384261131287,
|
| 7546 |
+
"learning_rate": 7.392583872866392e-05,
|
| 7547 |
+
"loss": 0.6819,
|
| 7548 |
+
"step": 1077
|
| 7549 |
+
},
|
| 7550 |
+
{
|
| 7551 |
+
"epoch": 1.2652582159624413,
|
| 7552 |
+
"grad_norm": 0.35997796058654785,
|
| 7553 |
+
"learning_rate": 7.380812242495586e-05,
|
| 7554 |
+
"loss": 0.6772,
|
| 7555 |
+
"step": 1078
|
| 7556 |
+
},
|
| 7557 |
+
{
|
| 7558 |
+
"epoch": 1.266431924882629,
|
| 7559 |
+
"grad_norm": 0.385193407535553,
|
| 7560 |
+
"learning_rate": 7.36904061212478e-05,
|
| 7561 |
+
"loss": 0.693,
|
| 7562 |
+
"step": 1079
|
| 7563 |
+
},
|
| 7564 |
+
{
|
| 7565 |
+
"epoch": 1.267605633802817,
|
| 7566 |
+
"grad_norm": 0.40055274963378906,
|
| 7567 |
+
"learning_rate": 7.357268981753973e-05,
|
| 7568 |
+
"loss": 0.6957,
|
| 7569 |
+
"step": 1080
|
| 7570 |
+
},
|
| 7571 |
+
{
|
| 7572 |
+
"epoch": 1.2687793427230047,
|
| 7573 |
+
"grad_norm": 0.35632067918777466,
|
| 7574 |
+
"learning_rate": 7.345497351383167e-05,
|
| 7575 |
+
"loss": 0.685,
|
| 7576 |
+
"step": 1081
|
| 7577 |
+
},
|
| 7578 |
+
{
|
| 7579 |
+
"epoch": 1.2699530516431925,
|
| 7580 |
+
"grad_norm": 0.3778478503227234,
|
| 7581 |
+
"learning_rate": 7.333725721012361e-05,
|
| 7582 |
+
"loss": 0.6657,
|
| 7583 |
+
"step": 1082
|
| 7584 |
+
},
|
| 7585 |
+
{
|
| 7586 |
+
"epoch": 1.2711267605633803,
|
| 7587 |
+
"grad_norm": 0.38718998432159424,
|
| 7588 |
+
"learning_rate": 7.321954090641554e-05,
|
| 7589 |
+
"loss": 0.6929,
|
| 7590 |
+
"step": 1083
|
| 7591 |
+
},
|
| 7592 |
+
{
|
| 7593 |
+
"epoch": 1.272300469483568,
|
| 7594 |
+
"grad_norm": 0.3769080936908722,
|
| 7595 |
+
"learning_rate": 7.310182460270748e-05,
|
| 7596 |
+
"loss": 0.7031,
|
| 7597 |
+
"step": 1084
|
| 7598 |
+
},
|
| 7599 |
+
{
|
| 7600 |
+
"epoch": 1.273474178403756,
|
| 7601 |
+
"grad_norm": 0.3617388904094696,
|
| 7602 |
+
"learning_rate": 7.298410829899941e-05,
|
| 7603 |
+
"loss": 0.665,
|
| 7604 |
+
"step": 1085
|
| 7605 |
+
},
|
| 7606 |
+
{
|
| 7607 |
+
"epoch": 1.2746478873239437,
|
| 7608 |
+
"grad_norm": 0.38341668248176575,
|
| 7609 |
+
"learning_rate": 7.286639199529135e-05,
|
| 7610 |
+
"loss": 0.6921,
|
| 7611 |
+
"step": 1086
|
| 7612 |
+
},
|
| 7613 |
+
{
|
| 7614 |
+
"epoch": 1.2758215962441315,
|
| 7615 |
+
"grad_norm": 0.36512598395347595,
|
| 7616 |
+
"learning_rate": 7.27486756915833e-05,
|
| 7617 |
+
"loss": 0.6582,
|
| 7618 |
+
"step": 1087
|
| 7619 |
+
},
|
| 7620 |
+
{
|
| 7621 |
+
"epoch": 1.2769953051643192,
|
| 7622 |
+
"grad_norm": 0.3690038025379181,
|
| 7623 |
+
"learning_rate": 7.263095938787522e-05,
|
| 7624 |
+
"loss": 0.6923,
|
| 7625 |
+
"step": 1088
|
| 7626 |
+
},
|
| 7627 |
+
{
|
| 7628 |
+
"epoch": 1.278169014084507,
|
| 7629 |
+
"grad_norm": 0.3716605007648468,
|
| 7630 |
+
"learning_rate": 7.251324308416716e-05,
|
| 7631 |
+
"loss": 0.6825,
|
| 7632 |
+
"step": 1089
|
| 7633 |
+
},
|
| 7634 |
+
{
|
| 7635 |
+
"epoch": 1.2793427230046948,
|
| 7636 |
+
"grad_norm": 0.3732469379901886,
|
| 7637 |
+
"learning_rate": 7.23955267804591e-05,
|
| 7638 |
+
"loss": 0.7064,
|
| 7639 |
+
"step": 1090
|
| 7640 |
+
},
|
| 7641 |
+
{
|
| 7642 |
+
"epoch": 1.2805164319248825,
|
| 7643 |
+
"grad_norm": 0.3982716500759125,
|
| 7644 |
+
"learning_rate": 7.227781047675103e-05,
|
| 7645 |
+
"loss": 0.7379,
|
| 7646 |
+
"step": 1091
|
| 7647 |
+
},
|
| 7648 |
+
{
|
| 7649 |
+
"epoch": 1.2816901408450705,
|
| 7650 |
+
"grad_norm": 0.36569979786872864,
|
| 7651 |
+
"learning_rate": 7.216009417304297e-05,
|
| 7652 |
+
"loss": 0.6309,
|
| 7653 |
+
"step": 1092
|
| 7654 |
+
},
|
| 7655 |
+
{
|
| 7656 |
+
"epoch": 1.2828638497652582,
|
| 7657 |
+
"grad_norm": 0.3752378523349762,
|
| 7658 |
+
"learning_rate": 7.204237786933492e-05,
|
| 7659 |
+
"loss": 0.655,
|
| 7660 |
+
"step": 1093
|
| 7661 |
+
},
|
| 7662 |
+
{
|
| 7663 |
+
"epoch": 1.284037558685446,
|
| 7664 |
+
"grad_norm": 0.3787980377674103,
|
| 7665 |
+
"learning_rate": 7.192466156562684e-05,
|
| 7666 |
+
"loss": 0.6503,
|
| 7667 |
+
"step": 1094
|
| 7668 |
+
},
|
| 7669 |
+
{
|
| 7670 |
+
"epoch": 1.2852112676056338,
|
| 7671 |
+
"grad_norm": 0.40723252296447754,
|
| 7672 |
+
"learning_rate": 7.180694526191878e-05,
|
| 7673 |
+
"loss": 0.6949,
|
| 7674 |
+
"step": 1095
|
| 7675 |
+
},
|
| 7676 |
+
{
|
| 7677 |
+
"epoch": 1.2863849765258215,
|
| 7678 |
+
"grad_norm": 0.36322662234306335,
|
| 7679 |
+
"learning_rate": 7.168922895821072e-05,
|
| 7680 |
+
"loss": 0.6924,
|
| 7681 |
+
"step": 1096
|
| 7682 |
+
},
|
| 7683 |
+
{
|
| 7684 |
+
"epoch": 1.2875586854460095,
|
| 7685 |
+
"grad_norm": 0.38535597920417786,
|
| 7686 |
+
"learning_rate": 7.157151265450265e-05,
|
| 7687 |
+
"loss": 0.6597,
|
| 7688 |
+
"step": 1097
|
| 7689 |
+
},
|
| 7690 |
+
{
|
| 7691 |
+
"epoch": 1.2887323943661972,
|
| 7692 |
+
"grad_norm": 0.3948807716369629,
|
| 7693 |
+
"learning_rate": 7.145379635079459e-05,
|
| 7694 |
+
"loss": 0.6594,
|
| 7695 |
+
"step": 1098
|
| 7696 |
+
},
|
| 7697 |
+
{
|
| 7698 |
+
"epoch": 1.289906103286385,
|
| 7699 |
+
"grad_norm": 0.3859221935272217,
|
| 7700 |
+
"learning_rate": 7.133608004708652e-05,
|
| 7701 |
+
"loss": 0.6894,
|
| 7702 |
+
"step": 1099
|
| 7703 |
+
},
|
| 7704 |
+
{
|
| 7705 |
+
"epoch": 1.2910798122065728,
|
| 7706 |
+
"grad_norm": 0.36739176511764526,
|
| 7707 |
+
"learning_rate": 7.121836374337846e-05,
|
| 7708 |
+
"loss": 0.6726,
|
| 7709 |
+
"step": 1100
|
| 7710 |
+
},
|
| 7711 |
+
{
|
| 7712 |
+
"epoch": 1.290653384119543,
|
| 7713 |
+
"grad_norm": 1.054391622543335,
|
| 7714 |
+
"learning_rate": 7.125220458553792e-05,
|
| 7715 |
+
"loss": 0.8774,
|
| 7716 |
+
"step": 1101
|
| 7717 |
+
},
|
| 7718 |
+
{
|
| 7719 |
+
"epoch": 1.2918253735716378,
|
| 7720 |
+
"grad_norm": 1.0680177211761475,
|
| 7721 |
+
"learning_rate": 7.113462669018225e-05,
|
| 7722 |
+
"loss": 0.8413,
|
| 7723 |
+
"step": 1102
|
| 7724 |
+
},
|
| 7725 |
+
{
|
| 7726 |
+
"epoch": 1.2929973630237328,
|
| 7727 |
+
"grad_norm": 0.6065592169761658,
|
| 7728 |
+
"learning_rate": 7.101704879482658e-05,
|
| 7729 |
+
"loss": 0.7806,
|
| 7730 |
+
"step": 1103
|
| 7731 |
+
},
|
| 7732 |
+
{
|
| 7733 |
+
"epoch": 1.2941693524758278,
|
| 7734 |
+
"grad_norm": 0.6377708315849304,
|
| 7735 |
+
"learning_rate": 7.089947089947089e-05,
|
| 7736 |
+
"loss": 0.7703,
|
| 7737 |
+
"step": 1104
|
| 7738 |
+
},
|
| 7739 |
+
{
|
| 7740 |
+
"epoch": 1.2953413419279227,
|
| 7741 |
+
"grad_norm": 0.522957980632782,
|
| 7742 |
+
"learning_rate": 7.078189300411523e-05,
|
| 7743 |
+
"loss": 0.7668,
|
| 7744 |
+
"step": 1105
|
| 7745 |
+
},
|
| 7746 |
+
{
|
| 7747 |
+
"epoch": 1.2965133313800177,
|
| 7748 |
+
"grad_norm": 0.5710256099700928,
|
| 7749 |
+
"learning_rate": 7.066431510875956e-05,
|
| 7750 |
+
"loss": 0.7143,
|
| 7751 |
+
"step": 1106
|
| 7752 |
+
},
|
| 7753 |
+
{
|
| 7754 |
+
"epoch": 1.2976853208321124,
|
| 7755 |
+
"grad_norm": 0.5131927728652954,
|
| 7756 |
+
"learning_rate": 7.054673721340387e-05,
|
| 7757 |
+
"loss": 0.7151,
|
| 7758 |
+
"step": 1107
|
| 7759 |
+
},
|
| 7760 |
+
{
|
| 7761 |
+
"epoch": 1.2988573102842074,
|
| 7762 |
+
"grad_norm": 0.495702862739563,
|
| 7763 |
+
"learning_rate": 7.042915931804821e-05,
|
| 7764 |
+
"loss": 0.7108,
|
| 7765 |
+
"step": 1108
|
| 7766 |
+
},
|
| 7767 |
+
{
|
| 7768 |
+
"epoch": 1.3000292997363023,
|
| 7769 |
+
"grad_norm": 0.49212130904197693,
|
| 7770 |
+
"learning_rate": 7.031158142269254e-05,
|
| 7771 |
+
"loss": 0.711,
|
| 7772 |
+
"step": 1109
|
| 7773 |
+
},
|
| 7774 |
+
{
|
| 7775 |
+
"epoch": 1.3012012891883973,
|
| 7776 |
+
"grad_norm": 0.4861961305141449,
|
| 7777 |
+
"learning_rate": 7.019400352733686e-05,
|
| 7778 |
+
"loss": 0.7562,
|
| 7779 |
+
"step": 1110
|
| 7780 |
+
},
|
| 7781 |
+
{
|
| 7782 |
+
"epoch": 1.3023732786404922,
|
| 7783 |
+
"grad_norm": 0.4575994610786438,
|
| 7784 |
+
"learning_rate": 7.00764256319812e-05,
|
| 7785 |
+
"loss": 0.7036,
|
| 7786 |
+
"step": 1111
|
| 7787 |
+
},
|
| 7788 |
+
{
|
| 7789 |
+
"epoch": 1.3035452680925872,
|
| 7790 |
+
"grad_norm": 0.4313446283340454,
|
| 7791 |
+
"learning_rate": 6.995884773662552e-05,
|
| 7792 |
+
"loss": 0.7124,
|
| 7793 |
+
"step": 1112
|
| 7794 |
+
},
|
| 7795 |
+
{
|
| 7796 |
+
"epoch": 1.3047172575446822,
|
| 7797 |
+
"grad_norm": 0.4154439866542816,
|
| 7798 |
+
"learning_rate": 6.984126984126984e-05,
|
| 7799 |
+
"loss": 0.6674,
|
| 7800 |
+
"step": 1113
|
| 7801 |
+
},
|
| 7802 |
+
{
|
| 7803 |
+
"epoch": 1.305889246996777,
|
| 7804 |
+
"grad_norm": 0.4371664226055145,
|
| 7805 |
+
"learning_rate": 6.972369194591418e-05,
|
| 7806 |
+
"loss": 0.7066,
|
| 7807 |
+
"step": 1114
|
| 7808 |
+
},
|
| 7809 |
+
{
|
| 7810 |
+
"epoch": 1.307061236448872,
|
| 7811 |
+
"grad_norm": 0.40978437662124634,
|
| 7812 |
+
"learning_rate": 6.96061140505585e-05,
|
| 7813 |
+
"loss": 0.6923,
|
| 7814 |
+
"step": 1115
|
| 7815 |
+
},
|
| 7816 |
+
{
|
| 7817 |
+
"epoch": 1.3082332259009668,
|
| 7818 |
+
"grad_norm": 0.39959609508514404,
|
| 7819 |
+
"learning_rate": 6.948853615520282e-05,
|
| 7820 |
+
"loss": 0.7224,
|
| 7821 |
+
"step": 1116
|
| 7822 |
+
},
|
| 7823 |
+
{
|
| 7824 |
+
"epoch": 1.3094052153530618,
|
| 7825 |
+
"grad_norm": 0.4188673794269562,
|
| 7826 |
+
"learning_rate": 6.937095825984715e-05,
|
| 7827 |
+
"loss": 0.727,
|
| 7828 |
+
"step": 1117
|
| 7829 |
+
},
|
| 7830 |
+
{
|
| 7831 |
+
"epoch": 1.3105772048051567,
|
| 7832 |
+
"grad_norm": 0.3936775028705597,
|
| 7833 |
+
"learning_rate": 6.925338036449149e-05,
|
| 7834 |
+
"loss": 0.6825,
|
| 7835 |
+
"step": 1118
|
| 7836 |
+
},
|
| 7837 |
+
{
|
| 7838 |
+
"epoch": 1.3117491942572517,
|
| 7839 |
+
"grad_norm": 0.39894506335258484,
|
| 7840 |
+
"learning_rate": 6.91358024691358e-05,
|
| 7841 |
+
"loss": 0.642,
|
| 7842 |
+
"step": 1119
|
| 7843 |
+
},
|
| 7844 |
+
{
|
| 7845 |
+
"epoch": 1.3129211837093466,
|
| 7846 |
+
"grad_norm": 0.4340292513370514,
|
| 7847 |
+
"learning_rate": 6.901822457378013e-05,
|
| 7848 |
+
"loss": 0.7259,
|
| 7849 |
+
"step": 1120
|
| 7850 |
+
},
|
| 7851 |
+
{
|
| 7852 |
+
"epoch": 1.3140931731614416,
|
| 7853 |
+
"grad_norm": 0.4102571904659271,
|
| 7854 |
+
"learning_rate": 6.890064667842447e-05,
|
| 7855 |
+
"loss": 0.6954,
|
| 7856 |
+
"step": 1121
|
| 7857 |
+
},
|
| 7858 |
+
{
|
| 7859 |
+
"epoch": 1.3152651626135365,
|
| 7860 |
+
"grad_norm": 0.4111669063568115,
|
| 7861 |
+
"learning_rate": 6.878306878306878e-05,
|
| 7862 |
+
"loss": 0.671,
|
| 7863 |
+
"step": 1122
|
| 7864 |
+
},
|
| 7865 |
+
{
|
| 7866 |
+
"epoch": 1.3164371520656313,
|
| 7867 |
+
"grad_norm": 0.4138496518135071,
|
| 7868 |
+
"learning_rate": 6.866549088771311e-05,
|
| 7869 |
+
"loss": 0.6387,
|
| 7870 |
+
"step": 1123
|
| 7871 |
+
},
|
| 7872 |
+
{
|
| 7873 |
+
"epoch": 1.3176091415177265,
|
| 7874 |
+
"grad_norm": 0.3867737054824829,
|
| 7875 |
+
"learning_rate": 6.854791299235745e-05,
|
| 7876 |
+
"loss": 0.6777,
|
| 7877 |
+
"step": 1124
|
| 7878 |
+
},
|
| 7879 |
+
{
|
| 7880 |
+
"epoch": 1.3187811309698212,
|
| 7881 |
+
"grad_norm": 0.40671873092651367,
|
| 7882 |
+
"learning_rate": 6.843033509700176e-05,
|
| 7883 |
+
"loss": 0.695,
|
| 7884 |
+
"step": 1125
|
| 7885 |
+
},
|
| 7886 |
+
{
|
| 7887 |
+
"epoch": 1.3199531204219161,
|
| 7888 |
+
"grad_norm": 0.3959422707557678,
|
| 7889 |
+
"learning_rate": 6.831275720164609e-05,
|
| 7890 |
+
"loss": 0.6755,
|
| 7891 |
+
"step": 1126
|
| 7892 |
+
},
|
| 7893 |
+
{
|
| 7894 |
+
"epoch": 1.321125109874011,
|
| 7895 |
+
"grad_norm": 0.3982046842575073,
|
| 7896 |
+
"learning_rate": 6.819517930629043e-05,
|
| 7897 |
+
"loss": 0.6898,
|
| 7898 |
+
"step": 1127
|
| 7899 |
+
},
|
| 7900 |
+
{
|
| 7901 |
+
"epoch": 1.322297099326106,
|
| 7902 |
+
"grad_norm": 0.3888896405696869,
|
| 7903 |
+
"learning_rate": 6.807760141093474e-05,
|
| 7904 |
+
"loss": 0.6595,
|
| 7905 |
+
"step": 1128
|
| 7906 |
+
},
|
| 7907 |
+
{
|
| 7908 |
+
"epoch": 1.323469088778201,
|
| 7909 |
+
"grad_norm": 0.4006032347679138,
|
| 7910 |
+
"learning_rate": 6.796002351557907e-05,
|
| 7911 |
+
"loss": 0.6882,
|
| 7912 |
+
"step": 1129
|
| 7913 |
+
},
|
| 7914 |
+
{
|
| 7915 |
+
"epoch": 1.324641078230296,
|
| 7916 |
+
"grad_norm": 0.37636417150497437,
|
| 7917 |
+
"learning_rate": 6.78424456202234e-05,
|
| 7918 |
+
"loss": 0.6689,
|
| 7919 |
+
"step": 1130
|
| 7920 |
+
},
|
| 7921 |
+
{
|
| 7922 |
+
"epoch": 1.325813067682391,
|
| 7923 |
+
"grad_norm": 0.3819728195667267,
|
| 7924 |
+
"learning_rate": 6.772486772486773e-05,
|
| 7925 |
+
"loss": 0.7133,
|
| 7926 |
+
"step": 1131
|
| 7927 |
+
},
|
| 7928 |
+
{
|
| 7929 |
+
"epoch": 1.3269850571344857,
|
| 7930 |
+
"grad_norm": 0.37200263142585754,
|
| 7931 |
+
"learning_rate": 6.760728982951205e-05,
|
| 7932 |
+
"loss": 0.6942,
|
| 7933 |
+
"step": 1132
|
| 7934 |
+
},
|
| 7935 |
+
{
|
| 7936 |
+
"epoch": 1.3281570465865808,
|
| 7937 |
+
"grad_norm": 0.3960074186325073,
|
| 7938 |
+
"learning_rate": 6.748971193415638e-05,
|
| 7939 |
+
"loss": 0.6271,
|
| 7940 |
+
"step": 1133
|
| 7941 |
+
},
|
| 7942 |
+
{
|
| 7943 |
+
"epoch": 1.3293290360386756,
|
| 7944 |
+
"grad_norm": 0.3755532205104828,
|
| 7945 |
+
"learning_rate": 6.737213403880071e-05,
|
| 7946 |
+
"loss": 0.6706,
|
| 7947 |
+
"step": 1134
|
| 7948 |
+
},
|
| 7949 |
+
{
|
| 7950 |
+
"epoch": 1.3305010254907705,
|
| 7951 |
+
"grad_norm": 0.387464314699173,
|
| 7952 |
+
"learning_rate": 6.725455614344503e-05,
|
| 7953 |
+
"loss": 0.6958,
|
| 7954 |
+
"step": 1135
|
| 7955 |
+
},
|
| 7956 |
+
{
|
| 7957 |
+
"epoch": 1.3316730149428655,
|
| 7958 |
+
"grad_norm": 0.3806421756744385,
|
| 7959 |
+
"learning_rate": 6.713697824808936e-05,
|
| 7960 |
+
"loss": 0.6928,
|
| 7961 |
+
"step": 1136
|
| 7962 |
+
},
|
| 7963 |
+
{
|
| 7964 |
+
"epoch": 1.3328450043949605,
|
| 7965 |
+
"grad_norm": 0.38345402479171753,
|
| 7966 |
+
"learning_rate": 6.701940035273369e-05,
|
| 7967 |
+
"loss": 0.7126,
|
| 7968 |
+
"step": 1137
|
| 7969 |
+
},
|
| 7970 |
+
{
|
| 7971 |
+
"epoch": 1.3340169938470554,
|
| 7972 |
+
"grad_norm": 0.3863559067249298,
|
| 7973 |
+
"learning_rate": 6.690182245737802e-05,
|
| 7974 |
+
"loss": 0.6914,
|
| 7975 |
+
"step": 1138
|
| 7976 |
+
},
|
| 7977 |
+
{
|
| 7978 |
+
"epoch": 1.3351889832991504,
|
| 7979 |
+
"grad_norm": 0.3891694247722626,
|
| 7980 |
+
"learning_rate": 6.678424456202234e-05,
|
| 7981 |
+
"loss": 0.6817,
|
| 7982 |
+
"step": 1139
|
| 7983 |
+
},
|
| 7984 |
+
{
|
| 7985 |
+
"epoch": 1.3363609727512453,
|
| 7986 |
+
"grad_norm": 0.4065246284008026,
|
| 7987 |
+
"learning_rate": 6.666666666666667e-05,
|
| 7988 |
+
"loss": 0.7028,
|
| 7989 |
+
"step": 1140
|
| 7990 |
+
},
|
| 7991 |
+
{
|
| 7992 |
+
"epoch": 1.33753296220334,
|
| 7993 |
+
"grad_norm": 0.38939380645751953,
|
| 7994 |
+
"learning_rate": 6.6549088771311e-05,
|
| 7995 |
+
"loss": 0.6528,
|
| 7996 |
+
"step": 1141
|
| 7997 |
+
},
|
| 7998 |
+
{
|
| 7999 |
+
"epoch": 1.3387049516554352,
|
| 8000 |
+
"grad_norm": 0.3831149935722351,
|
| 8001 |
+
"learning_rate": 6.643151087595532e-05,
|
| 8002 |
+
"loss": 0.675,
|
| 8003 |
+
"step": 1142
|
| 8004 |
+
},
|
| 8005 |
+
{
|
| 8006 |
+
"epoch": 1.33987694110753,
|
| 8007 |
+
"grad_norm": 0.4192376434803009,
|
| 8008 |
+
"learning_rate": 6.631393298059965e-05,
|
| 8009 |
+
"loss": 0.7131,
|
| 8010 |
+
"step": 1143
|
| 8011 |
+
},
|
| 8012 |
+
{
|
| 8013 |
+
"epoch": 1.341048930559625,
|
| 8014 |
+
"grad_norm": 0.38939863443374634,
|
| 8015 |
+
"learning_rate": 6.619635508524398e-05,
|
| 8016 |
+
"loss": 0.6696,
|
| 8017 |
+
"step": 1144
|
| 8018 |
+
},
|
| 8019 |
+
{
|
| 8020 |
+
"epoch": 1.3422209200117199,
|
| 8021 |
+
"grad_norm": 0.39097532629966736,
|
| 8022 |
+
"learning_rate": 6.60787771898883e-05,
|
| 8023 |
+
"loss": 0.6845,
|
| 8024 |
+
"step": 1145
|
| 8025 |
+
},
|
| 8026 |
+
{
|
| 8027 |
+
"epoch": 1.3433929094638148,
|
| 8028 |
+
"grad_norm": 0.39098918437957764,
|
| 8029 |
+
"learning_rate": 6.596119929453263e-05,
|
| 8030 |
+
"loss": 0.6643,
|
| 8031 |
+
"step": 1146
|
| 8032 |
+
},
|
| 8033 |
+
{
|
| 8034 |
+
"epoch": 1.3445648989159098,
|
| 8035 |
+
"grad_norm": 0.41722220182418823,
|
| 8036 |
+
"learning_rate": 6.584362139917696e-05,
|
| 8037 |
+
"loss": 0.6461,
|
| 8038 |
+
"step": 1147
|
| 8039 |
+
},
|
| 8040 |
+
{
|
| 8041 |
+
"epoch": 1.3457368883680048,
|
| 8042 |
+
"grad_norm": 0.40393978357315063,
|
| 8043 |
+
"learning_rate": 6.572604350382129e-05,
|
| 8044 |
+
"loss": 0.677,
|
| 8045 |
+
"step": 1148
|
| 8046 |
+
},
|
| 8047 |
+
{
|
| 8048 |
+
"epoch": 1.3469088778200997,
|
| 8049 |
+
"grad_norm": 0.37448999285697937,
|
| 8050 |
+
"learning_rate": 6.560846560846561e-05,
|
| 8051 |
+
"loss": 0.6534,
|
| 8052 |
+
"step": 1149
|
| 8053 |
+
},
|
| 8054 |
+
{
|
| 8055 |
+
"epoch": 1.3480808672721944,
|
| 8056 |
+
"grad_norm": 0.3823452889919281,
|
| 8057 |
+
"learning_rate": 6.549088771310994e-05,
|
| 8058 |
+
"loss": 0.6269,
|
| 8059 |
+
"step": 1150
|
| 8060 |
+
},
|
| 8061 |
+
{
|
| 8062 |
+
"epoch": 1.3492528567242894,
|
| 8063 |
+
"grad_norm": 0.4128420650959015,
|
| 8064 |
+
"learning_rate": 6.537330981775427e-05,
|
| 8065 |
+
"loss": 0.6659,
|
| 8066 |
+
"step": 1151
|
| 8067 |
+
},
|
| 8068 |
+
{
|
| 8069 |
+
"epoch": 1.3504248461763844,
|
| 8070 |
+
"grad_norm": 0.385935515165329,
|
| 8071 |
+
"learning_rate": 6.525573192239858e-05,
|
| 8072 |
+
"loss": 0.6501,
|
| 8073 |
+
"step": 1152
|
| 8074 |
+
},
|
| 8075 |
+
{
|
| 8076 |
+
"epoch": 1.3515968356284793,
|
| 8077 |
+
"grad_norm": 0.39446333050727844,
|
| 8078 |
+
"learning_rate": 6.513815402704292e-05,
|
| 8079 |
+
"loss": 0.6607,
|
| 8080 |
+
"step": 1153
|
| 8081 |
+
},
|
| 8082 |
+
{
|
| 8083 |
+
"epoch": 1.3527688250805743,
|
| 8084 |
+
"grad_norm": 0.39609313011169434,
|
| 8085 |
+
"learning_rate": 6.502057613168725e-05,
|
| 8086 |
+
"loss": 0.6408,
|
| 8087 |
+
"step": 1154
|
| 8088 |
+
},
|
| 8089 |
+
{
|
| 8090 |
+
"epoch": 1.3539408145326692,
|
| 8091 |
+
"grad_norm": 0.4118976891040802,
|
| 8092 |
+
"learning_rate": 6.490299823633156e-05,
|
| 8093 |
+
"loss": 0.6783,
|
| 8094 |
+
"step": 1155
|
| 8095 |
+
},
|
| 8096 |
+
{
|
| 8097 |
+
"epoch": 1.3551128039847642,
|
| 8098 |
+
"grad_norm": 0.39040106534957886,
|
| 8099 |
+
"learning_rate": 6.47854203409759e-05,
|
| 8100 |
+
"loss": 0.6996,
|
| 8101 |
+
"step": 1156
|
| 8102 |
+
},
|
| 8103 |
+
{
|
| 8104 |
+
"epoch": 1.3562847934368591,
|
| 8105 |
+
"grad_norm": 0.3837215304374695,
|
| 8106 |
+
"learning_rate": 6.466784244562023e-05,
|
| 8107 |
+
"loss": 0.6536,
|
| 8108 |
+
"step": 1157
|
| 8109 |
+
},
|
| 8110 |
+
{
|
| 8111 |
+
"epoch": 1.357456782888954,
|
| 8112 |
+
"grad_norm": 0.38842037320137024,
|
| 8113 |
+
"learning_rate": 6.455026455026454e-05,
|
| 8114 |
+
"loss": 0.675,
|
| 8115 |
+
"step": 1158
|
| 8116 |
+
},
|
| 8117 |
+
{
|
| 8118 |
+
"epoch": 1.3586287723410488,
|
| 8119 |
+
"grad_norm": 0.3818652927875519,
|
| 8120 |
+
"learning_rate": 6.443268665490889e-05,
|
| 8121 |
+
"loss": 0.6821,
|
| 8122 |
+
"step": 1159
|
| 8123 |
+
},
|
| 8124 |
+
{
|
| 8125 |
+
"epoch": 1.3598007617931438,
|
| 8126 |
+
"grad_norm": 0.38457590341567993,
|
| 8127 |
+
"learning_rate": 6.431510875955321e-05,
|
| 8128 |
+
"loss": 0.6801,
|
| 8129 |
+
"step": 1160
|
| 8130 |
+
},
|
| 8131 |
+
{
|
| 8132 |
+
"epoch": 1.3609727512452388,
|
| 8133 |
+
"grad_norm": 0.40308085083961487,
|
| 8134 |
+
"learning_rate": 6.419753086419753e-05,
|
| 8135 |
+
"loss": 0.7282,
|
| 8136 |
+
"step": 1161
|
| 8137 |
+
},
|
| 8138 |
+
{
|
| 8139 |
+
"epoch": 1.3621447406973337,
|
| 8140 |
+
"grad_norm": 0.3923408091068268,
|
| 8141 |
+
"learning_rate": 6.407995296884187e-05,
|
| 8142 |
+
"loss": 0.6771,
|
| 8143 |
+
"step": 1162
|
| 8144 |
+
},
|
| 8145 |
+
{
|
| 8146 |
+
"epoch": 1.3633167301494287,
|
| 8147 |
+
"grad_norm": 0.35983577370643616,
|
| 8148 |
+
"learning_rate": 6.39623750734862e-05,
|
| 8149 |
+
"loss": 0.6684,
|
| 8150 |
+
"step": 1163
|
| 8151 |
+
},
|
| 8152 |
+
{
|
| 8153 |
+
"epoch": 1.3644887196015236,
|
| 8154 |
+
"grad_norm": 0.38956505060195923,
|
| 8155 |
+
"learning_rate": 6.384479717813051e-05,
|
| 8156 |
+
"loss": 0.6456,
|
| 8157 |
+
"step": 1164
|
| 8158 |
+
},
|
| 8159 |
+
{
|
| 8160 |
+
"epoch": 1.3656607090536186,
|
| 8161 |
+
"grad_norm": 0.3745352327823639,
|
| 8162 |
+
"learning_rate": 6.372721928277484e-05,
|
| 8163 |
+
"loss": 0.6628,
|
| 8164 |
+
"step": 1165
|
| 8165 |
+
},
|
| 8166 |
+
{
|
| 8167 |
+
"epoch": 1.3668326985057133,
|
| 8168 |
+
"grad_norm": 0.3648051619529724,
|
| 8169 |
+
"learning_rate": 6.360964138741918e-05,
|
| 8170 |
+
"loss": 0.6341,
|
| 8171 |
+
"step": 1166
|
| 8172 |
+
},
|
| 8173 |
+
{
|
| 8174 |
+
"epoch": 1.3680046879578085,
|
| 8175 |
+
"grad_norm": 0.3933015763759613,
|
| 8176 |
+
"learning_rate": 6.349206349206349e-05,
|
| 8177 |
+
"loss": 0.6862,
|
| 8178 |
+
"step": 1167
|
| 8179 |
+
},
|
| 8180 |
+
{
|
| 8181 |
+
"epoch": 1.3691766774099032,
|
| 8182 |
+
"grad_norm": 0.3905039429664612,
|
| 8183 |
+
"learning_rate": 6.337448559670782e-05,
|
| 8184 |
+
"loss": 0.6697,
|
| 8185 |
+
"step": 1168
|
| 8186 |
+
},
|
| 8187 |
+
{
|
| 8188 |
+
"epoch": 1.3703486668619982,
|
| 8189 |
+
"grad_norm": 0.39760053157806396,
|
| 8190 |
+
"learning_rate": 6.325690770135216e-05,
|
| 8191 |
+
"loss": 0.6903,
|
| 8192 |
+
"step": 1169
|
| 8193 |
+
},
|
| 8194 |
+
{
|
| 8195 |
+
"epoch": 1.3715206563140931,
|
| 8196 |
+
"grad_norm": 0.38109833002090454,
|
| 8197 |
+
"learning_rate": 6.313932980599647e-05,
|
| 8198 |
+
"loss": 0.6539,
|
| 8199 |
+
"step": 1170
|
| 8200 |
+
},
|
| 8201 |
+
{
|
| 8202 |
+
"epoch": 1.372692645766188,
|
| 8203 |
+
"grad_norm": 0.3650861978530884,
|
| 8204 |
+
"learning_rate": 6.30217519106408e-05,
|
| 8205 |
+
"loss": 0.6749,
|
| 8206 |
+
"step": 1171
|
| 8207 |
+
},
|
| 8208 |
+
{
|
| 8209 |
+
"epoch": 1.373864635218283,
|
| 8210 |
+
"grad_norm": 0.35274556279182434,
|
| 8211 |
+
"learning_rate": 6.290417401528514e-05,
|
| 8212 |
+
"loss": 0.6421,
|
| 8213 |
+
"step": 1172
|
| 8214 |
+
},
|
| 8215 |
+
{
|
| 8216 |
+
"epoch": 1.375036624670378,
|
| 8217 |
+
"grad_norm": 0.41130268573760986,
|
| 8218 |
+
"learning_rate": 6.278659611992945e-05,
|
| 8219 |
+
"loss": 0.7218,
|
| 8220 |
+
"step": 1173
|
| 8221 |
+
},
|
| 8222 |
+
{
|
| 8223 |
+
"epoch": 1.376208614122473,
|
| 8224 |
+
"grad_norm": 0.3993924856185913,
|
| 8225 |
+
"learning_rate": 6.266901822457378e-05,
|
| 8226 |
+
"loss": 0.675,
|
| 8227 |
+
"step": 1174
|
| 8228 |
+
},
|
| 8229 |
+
{
|
| 8230 |
+
"epoch": 1.3773806035745677,
|
| 8231 |
+
"grad_norm": 0.3782753348350525,
|
| 8232 |
+
"learning_rate": 6.255144032921812e-05,
|
| 8233 |
+
"loss": 0.6316,
|
| 8234 |
+
"step": 1175
|
| 8235 |
+
},
|
| 8236 |
+
{
|
| 8237 |
+
"epoch": 1.3785525930266629,
|
| 8238 |
+
"grad_norm": 0.3624318242073059,
|
| 8239 |
+
"learning_rate": 6.243386243386243e-05,
|
| 8240 |
+
"loss": 0.65,
|
| 8241 |
+
"step": 1176
|
| 8242 |
+
},
|
| 8243 |
+
{
|
| 8244 |
+
"epoch": 1.3797245824787576,
|
| 8245 |
+
"grad_norm": 0.3888125717639923,
|
| 8246 |
+
"learning_rate": 6.231628453850676e-05,
|
| 8247 |
+
"loss": 0.7022,
|
| 8248 |
+
"step": 1177
|
| 8249 |
+
},
|
| 8250 |
+
{
|
| 8251 |
+
"epoch": 1.3808965719308526,
|
| 8252 |
+
"grad_norm": 0.3882753252983093,
|
| 8253 |
+
"learning_rate": 6.219870664315109e-05,
|
| 8254 |
+
"loss": 0.6866,
|
| 8255 |
+
"step": 1178
|
| 8256 |
+
},
|
| 8257 |
+
{
|
| 8258 |
+
"epoch": 1.3820685613829475,
|
| 8259 |
+
"grad_norm": 0.39087679982185364,
|
| 8260 |
+
"learning_rate": 6.208112874779542e-05,
|
| 8261 |
+
"loss": 0.6398,
|
| 8262 |
+
"step": 1179
|
| 8263 |
+
},
|
| 8264 |
+
{
|
| 8265 |
+
"epoch": 1.3832405508350425,
|
| 8266 |
+
"grad_norm": 0.40074875950813293,
|
| 8267 |
+
"learning_rate": 6.196355085243974e-05,
|
| 8268 |
+
"loss": 0.6446,
|
| 8269 |
+
"step": 1180
|
| 8270 |
+
},
|
| 8271 |
+
{
|
| 8272 |
+
"epoch": 1.3844125402871374,
|
| 8273 |
+
"grad_norm": 0.40180960297584534,
|
| 8274 |
+
"learning_rate": 6.184597295708407e-05,
|
| 8275 |
+
"loss": 0.6994,
|
| 8276 |
+
"step": 1181
|
| 8277 |
+
},
|
| 8278 |
+
{
|
| 8279 |
+
"epoch": 1.3855845297392324,
|
| 8280 |
+
"grad_norm": 0.38567396998405457,
|
| 8281 |
+
"learning_rate": 6.17283950617284e-05,
|
| 8282 |
+
"loss": 0.6659,
|
| 8283 |
+
"step": 1182
|
| 8284 |
+
},
|
| 8285 |
+
{
|
| 8286 |
+
"epoch": 1.3867565191913274,
|
| 8287 |
+
"grad_norm": 0.3690516948699951,
|
| 8288 |
+
"learning_rate": 6.161081716637272e-05,
|
| 8289 |
+
"loss": 0.6203,
|
| 8290 |
+
"step": 1183
|
| 8291 |
+
},
|
| 8292 |
+
{
|
| 8293 |
+
"epoch": 1.387928508643422,
|
| 8294 |
+
"grad_norm": 0.4022299349308014,
|
| 8295 |
+
"learning_rate": 6.149323927101705e-05,
|
| 8296 |
+
"loss": 0.6684,
|
| 8297 |
+
"step": 1184
|
| 8298 |
+
},
|
| 8299 |
+
{
|
| 8300 |
+
"epoch": 1.3891004980955173,
|
| 8301 |
+
"grad_norm": 0.384766161441803,
|
| 8302 |
+
"learning_rate": 6.137566137566138e-05,
|
| 8303 |
+
"loss": 0.6941,
|
| 8304 |
+
"step": 1185
|
| 8305 |
+
},
|
| 8306 |
+
{
|
| 8307 |
+
"epoch": 1.390272487547612,
|
| 8308 |
+
"grad_norm": 0.3742985427379608,
|
| 8309 |
+
"learning_rate": 6.12580834803057e-05,
|
| 8310 |
+
"loss": 0.6947,
|
| 8311 |
+
"step": 1186
|
| 8312 |
+
},
|
| 8313 |
+
{
|
| 8314 |
+
"epoch": 1.391444476999707,
|
| 8315 |
+
"grad_norm": 0.40857595205307007,
|
| 8316 |
+
"learning_rate": 6.114050558495003e-05,
|
| 8317 |
+
"loss": 0.6637,
|
| 8318 |
+
"step": 1187
|
| 8319 |
+
},
|
| 8320 |
+
{
|
| 8321 |
+
"epoch": 1.392616466451802,
|
| 8322 |
+
"grad_norm": 0.3824119567871094,
|
| 8323 |
+
"learning_rate": 6.102292768959435e-05,
|
| 8324 |
+
"loss": 0.6357,
|
| 8325 |
+
"step": 1188
|
| 8326 |
+
},
|
| 8327 |
+
{
|
| 8328 |
+
"epoch": 1.3937884559038969,
|
| 8329 |
+
"grad_norm": 0.39211392402648926,
|
| 8330 |
+
"learning_rate": 6.0905349794238687e-05,
|
| 8331 |
+
"loss": 0.644,
|
| 8332 |
+
"step": 1189
|
| 8333 |
+
},
|
| 8334 |
+
{
|
| 8335 |
+
"epoch": 1.3949604453559918,
|
| 8336 |
+
"grad_norm": 0.36606866121292114,
|
| 8337 |
+
"learning_rate": 6.0787771898883014e-05,
|
| 8338 |
+
"loss": 0.6388,
|
| 8339 |
+
"step": 1190
|
| 8340 |
+
},
|
| 8341 |
+
{
|
| 8342 |
+
"epoch": 1.3961324348080868,
|
| 8343 |
+
"grad_norm": 0.38364294171333313,
|
| 8344 |
+
"learning_rate": 6.0670194003527334e-05,
|
| 8345 |
+
"loss": 0.6561,
|
| 8346 |
+
"step": 1191
|
| 8347 |
+
},
|
| 8348 |
+
{
|
| 8349 |
+
"epoch": 1.3973044242601818,
|
| 8350 |
+
"grad_norm": 0.377848356962204,
|
| 8351 |
+
"learning_rate": 6.055261610817167e-05,
|
| 8352 |
+
"loss": 0.6916,
|
| 8353 |
+
"step": 1192
|
| 8354 |
+
},
|
| 8355 |
+
{
|
| 8356 |
+
"epoch": 1.3984764137122765,
|
| 8357 |
+
"grad_norm": 0.413018137216568,
|
| 8358 |
+
"learning_rate": 6.0435038212815995e-05,
|
| 8359 |
+
"loss": 0.6752,
|
| 8360 |
+
"step": 1193
|
| 8361 |
+
},
|
| 8362 |
+
{
|
| 8363 |
+
"epoch": 1.3996484031643714,
|
| 8364 |
+
"grad_norm": 0.3800462782382965,
|
| 8365 |
+
"learning_rate": 6.0317460317460316e-05,
|
| 8366 |
+
"loss": 0.6916,
|
| 8367 |
+
"step": 1194
|
| 8368 |
+
},
|
| 8369 |
+
{
|
| 8370 |
+
"epoch": 1.4008203926164664,
|
| 8371 |
+
"grad_norm": 0.38262689113616943,
|
| 8372 |
+
"learning_rate": 6.019988242210465e-05,
|
| 8373 |
+
"loss": 0.632,
|
| 8374 |
+
"step": 1195
|
| 8375 |
+
},
|
| 8376 |
+
{
|
| 8377 |
+
"epoch": 1.4019923820685614,
|
| 8378 |
+
"grad_norm": 0.37755706906318665,
|
| 8379 |
+
"learning_rate": 6.0082304526748977e-05,
|
| 8380 |
+
"loss": 0.608,
|
| 8381 |
+
"step": 1196
|
| 8382 |
+
},
|
| 8383 |
+
{
|
| 8384 |
+
"epoch": 1.4031643715206563,
|
| 8385 |
+
"grad_norm": 0.4054010510444641,
|
| 8386 |
+
"learning_rate": 5.99647266313933e-05,
|
| 8387 |
+
"loss": 0.6952,
|
| 8388 |
+
"step": 1197
|
| 8389 |
+
},
|
| 8390 |
+
{
|
| 8391 |
+
"epoch": 1.4043363609727513,
|
| 8392 |
+
"grad_norm": 0.38393253087997437,
|
| 8393 |
+
"learning_rate": 5.984714873603763e-05,
|
| 8394 |
+
"loss": 0.6742,
|
| 8395 |
+
"step": 1198
|
| 8396 |
+
},
|
| 8397 |
+
{
|
| 8398 |
+
"epoch": 1.4055083504248462,
|
| 8399 |
+
"grad_norm": 0.3715971112251282,
|
| 8400 |
+
"learning_rate": 5.972957084068196e-05,
|
| 8401 |
+
"loss": 0.6944,
|
| 8402 |
+
"step": 1199
|
| 8403 |
+
},
|
| 8404 |
+
{
|
| 8405 |
+
"epoch": 1.4066803398769412,
|
| 8406 |
+
"grad_norm": 0.3809766471385956,
|
| 8407 |
+
"learning_rate": 5.961199294532628e-05,
|
| 8408 |
+
"loss": 0.69,
|
| 8409 |
+
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 1.4078523293290361,
|
| 8413 |
+
"grad_norm": 0.42656511068344116,
|
| 8414 |
+
"learning_rate": 5.9494415049970606e-05,
|
| 8415 |
+
"loss": 0.6761,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 1.4090243187811309,
|
| 8420 |
+
"grad_norm": 0.39485815167427063,
|
| 8421 |
+
"learning_rate": 5.937683715461494e-05,
|
| 8422 |
+
"loss": 0.6581,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 1.4101963082332258,
|
| 8427 |
+
"grad_norm": 0.39925625920295715,
|
| 8428 |
+
"learning_rate": 5.925925925925926e-05,
|
| 8429 |
+
"loss": 0.6785,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 1.4113682976853208,
|
| 8434 |
+
"grad_norm": 0.38978663086891174,
|
| 8435 |
+
"learning_rate": 5.914168136390359e-05,
|
| 8436 |
+
"loss": 0.6814,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 1.4125402871374158,
|
| 8441 |
+
"grad_norm": 0.37807661294937134,
|
| 8442 |
+
"learning_rate": 5.902410346854792e-05,
|
| 8443 |
+
"loss": 0.6872,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 1.4137122765895107,
|
| 8448 |
+
"grad_norm": 0.3912755846977234,
|
| 8449 |
+
"learning_rate": 5.890652557319224e-05,
|
| 8450 |
+
"loss": 0.6343,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 1.4148842660416057,
|
| 8455 |
+
"grad_norm": 0.39116203784942627,
|
| 8456 |
+
"learning_rate": 5.878894767783657e-05,
|
| 8457 |
+
"loss": 0.63,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 1.4160562554937006,
|
| 8462 |
+
"grad_norm": 0.4052892029285431,
|
| 8463 |
+
"learning_rate": 5.86713697824809e-05,
|
| 8464 |
+
"loss": 0.6698,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 1.4172282449457954,
|
| 8469 |
+
"grad_norm": 0.3755436837673187,
|
| 8470 |
+
"learning_rate": 5.855379188712522e-05,
|
| 8471 |
+
"loss": 0.6966,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 1.4184002343978905,
|
| 8476 |
+
"grad_norm": 0.37595874071121216,
|
| 8477 |
+
"learning_rate": 5.843621399176955e-05,
|
| 8478 |
+
"loss": 0.6879,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 1.4195722238499853,
|
| 8483 |
+
"grad_norm": 0.4029812514781952,
|
| 8484 |
+
"learning_rate": 5.8318636096413884e-05,
|
| 8485 |
+
"loss": 0.6469,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 1.4207442133020802,
|
| 8490 |
+
"grad_norm": 0.37568360567092896,
|
| 8491 |
+
"learning_rate": 5.82010582010582e-05,
|
| 8492 |
+
"loss": 0.663,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 1.4219162027541752,
|
| 8497 |
+
"grad_norm": 0.38563036918640137,
|
| 8498 |
+
"learning_rate": 5.808348030570253e-05,
|
| 8499 |
+
"loss": 0.6849,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 1.4230881922062701,
|
| 8504 |
+
"grad_norm": 0.3693836033344269,
|
| 8505 |
+
"learning_rate": 5.796590241034686e-05,
|
| 8506 |
+
"loss": 0.6502,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 1.424260181658365,
|
| 8511 |
+
"grad_norm": 0.3790949583053589,
|
| 8512 |
+
"learning_rate": 5.784832451499118e-05,
|
| 8513 |
+
"loss": 0.6848,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 1.42543217111046,
|
| 8518 |
+
"grad_norm": 0.379332572221756,
|
| 8519 |
+
"learning_rate": 5.773074661963551e-05,
|
| 8520 |
+
"loss": 0.642,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 1.426604160562555,
|
| 8525 |
+
"grad_norm": 0.38053902983665466,
|
| 8526 |
+
"learning_rate": 5.761316872427984e-05,
|
| 8527 |
+
"loss": 0.7017,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 1.4277761500146497,
|
| 8532 |
+
"grad_norm": 0.39065247774124146,
|
| 8533 |
+
"learning_rate": 5.749559082892416e-05,
|
| 8534 |
+
"loss": 0.6959,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 1.428948139466745,
|
| 8539 |
+
"grad_norm": 0.40035319328308105,
|
| 8540 |
+
"learning_rate": 5.7378012933568494e-05,
|
| 8541 |
+
"loss": 0.6378,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 1.4301201289188397,
|
| 8546 |
+
"grad_norm": 0.44022828340530396,
|
| 8547 |
+
"learning_rate": 5.726043503821282e-05,
|
| 8548 |
+
"loss": 0.6805,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 1.4312921183709346,
|
| 8553 |
+
"grad_norm": 0.37635087966918945,
|
| 8554 |
+
"learning_rate": 5.714285714285714e-05,
|
| 8555 |
+
"loss": 0.6437,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 1.4324641078230296,
|
| 8560 |
+
"grad_norm": 0.3882914185523987,
|
| 8561 |
+
"learning_rate": 5.7025279247501476e-05,
|
| 8562 |
+
"loss": 0.6468,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 1.4336360972751245,
|
| 8567 |
+
"grad_norm": 0.3925718665122986,
|
| 8568 |
+
"learning_rate": 5.69077013521458e-05,
|
| 8569 |
+
"loss": 0.6516,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 1.4348080867272195,
|
| 8574 |
+
"grad_norm": 0.37410497665405273,
|
| 8575 |
+
"learning_rate": 5.679012345679012e-05,
|
| 8576 |
+
"loss": 0.6963,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 1.4359800761793144,
|
| 8581 |
+
"grad_norm": 0.42080360651016235,
|
| 8582 |
+
"learning_rate": 5.667254556143445e-05,
|
| 8583 |
+
"loss": 0.6386,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 1.4371520656314094,
|
| 8588 |
+
"grad_norm": 0.41487985849380493,
|
| 8589 |
+
"learning_rate": 5.6554967666078784e-05,
|
| 8590 |
+
"loss": 0.6914,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 1.4383240550835041,
|
| 8595 |
+
"grad_norm": 0.36109688878059387,
|
| 8596 |
+
"learning_rate": 5.6437389770723105e-05,
|
| 8597 |
+
"loss": 0.6672,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 1.4394960445355993,
|
| 8602 |
+
"grad_norm": 0.37526097893714905,
|
| 8603 |
+
"learning_rate": 5.631981187536743e-05,
|
| 8604 |
+
"loss": 0.6437,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 1.440668033987694,
|
| 8609 |
+
"grad_norm": 0.38677671551704407,
|
| 8610 |
+
"learning_rate": 5.6202233980011766e-05,
|
| 8611 |
+
"loss": 0.6432,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 1.441840023439789,
|
| 8616 |
+
"grad_norm": 0.37004703283309937,
|
| 8617 |
+
"learning_rate": 5.6084656084656086e-05,
|
| 8618 |
+
"loss": 0.6873,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 1.443012012891884,
|
| 8623 |
+
"grad_norm": 0.363845556974411,
|
| 8624 |
+
"learning_rate": 5.596707818930041e-05,
|
| 8625 |
+
"loss": 0.6388,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 1.444184002343979,
|
| 8630 |
+
"grad_norm": 0.3900223970413208,
|
| 8631 |
+
"learning_rate": 5.584950029394475e-05,
|
| 8632 |
+
"loss": 0.636,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 1.4453559917960739,
|
| 8637 |
+
"grad_norm": 0.37807023525238037,
|
| 8638 |
+
"learning_rate": 5.573192239858907e-05,
|
| 8639 |
+
"loss": 0.6699,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 1.4465279812481688,
|
| 8644 |
+
"grad_norm": 0.3589508533477783,
|
| 8645 |
+
"learning_rate": 5.5614344503233395e-05,
|
| 8646 |
+
"loss": 0.6256,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 1.4476999707002638,
|
| 8651 |
+
"grad_norm": 0.3793289065361023,
|
| 8652 |
+
"learning_rate": 5.549676660787773e-05,
|
| 8653 |
+
"loss": 0.66,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 1.4488719601523585,
|
| 8658 |
+
"grad_norm": 0.38588500022888184,
|
| 8659 |
+
"learning_rate": 5.537918871252204e-05,
|
| 8660 |
+
"loss": 0.6795,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 1.4500439496044535,
|
| 8665 |
+
"grad_norm": 0.3903558552265167,
|
| 8666 |
+
"learning_rate": 5.5261610817166376e-05,
|
| 8667 |
+
"loss": 0.6871,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 1.4512159390565484,
|
| 8672 |
+
"grad_norm": 0.3721795082092285,
|
| 8673 |
+
"learning_rate": 5.51440329218107e-05,
|
| 8674 |
+
"loss": 0.6524,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 1.4523879285086434,
|
| 8679 |
+
"grad_norm": 0.41447609663009644,
|
| 8680 |
+
"learning_rate": 5.5026455026455024e-05,
|
| 8681 |
+
"loss": 0.6711,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 1.4535599179607384,
|
| 8686 |
+
"grad_norm": 0.3805668354034424,
|
| 8687 |
+
"learning_rate": 5.490887713109936e-05,
|
| 8688 |
+
"loss": 0.6467,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 1.4547319074128333,
|
| 8693 |
+
"grad_norm": 0.3742830157279968,
|
| 8694 |
+
"learning_rate": 5.4791299235743685e-05,
|
| 8695 |
+
"loss": 0.6497,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 1.4559038968649283,
|
| 8700 |
+
"grad_norm": 0.3923674523830414,
|
| 8701 |
+
"learning_rate": 5.4673721340388005e-05,
|
| 8702 |
+
"loss": 0.6446,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 1.4570758863170232,
|
| 8707 |
+
"grad_norm": 0.4189533591270447,
|
| 8708 |
+
"learning_rate": 5.455614344503234e-05,
|
| 8709 |
+
"loss": 0.688,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 1.4582478757691182,
|
| 8714 |
+
"grad_norm": 0.38281819224357605,
|
| 8715 |
+
"learning_rate": 5.4438565549676666e-05,
|
| 8716 |
+
"loss": 0.6543,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 1.459419865221213,
|
| 8721 |
+
"grad_norm": 0.3858702480792999,
|
| 8722 |
+
"learning_rate": 5.4320987654320986e-05,
|
| 8723 |
+
"loss": 0.6822,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 1.4605918546733079,
|
| 8728 |
+
"grad_norm": 0.41707929968833923,
|
| 8729 |
+
"learning_rate": 5.420340975896532e-05,
|
| 8730 |
+
"loss": 0.6261,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 1.4617638441254028,
|
| 8735 |
+
"grad_norm": 0.36858686804771423,
|
| 8736 |
+
"learning_rate": 5.408583186360965e-05,
|
| 8737 |
+
"loss": 0.6457,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 1.4629358335774978,
|
| 8742 |
+
"grad_norm": 0.4082486033439636,
|
| 8743 |
+
"learning_rate": 5.396825396825397e-05,
|
| 8744 |
+
"loss": 0.6932,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 1.4641078230295927,
|
| 8749 |
+
"grad_norm": 0.3880460858345032,
|
| 8750 |
+
"learning_rate": 5.3850676072898295e-05,
|
| 8751 |
+
"loss": 0.6646,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 1.4652798124816877,
|
| 8756 |
+
"grad_norm": 0.3867412209510803,
|
| 8757 |
+
"learning_rate": 5.373309817754263e-05,
|
| 8758 |
+
"loss": 0.7426,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 1.4664518019337827,
|
| 8763 |
+
"grad_norm": 0.3632020354270935,
|
| 8764 |
+
"learning_rate": 5.361552028218695e-05,
|
| 8765 |
+
"loss": 0.6446,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 1.4676237913858774,
|
| 8770 |
+
"grad_norm": 0.36393246054649353,
|
| 8771 |
+
"learning_rate": 5.3497942386831277e-05,
|
| 8772 |
+
"loss": 0.6056,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 1.4687957808379726,
|
| 8777 |
+
"grad_norm": 0.37027665972709656,
|
| 8778 |
+
"learning_rate": 5.338036449147561e-05,
|
| 8779 |
+
"loss": 0.672,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 1.4699677702900673,
|
| 8784 |
+
"grad_norm": 0.35585132241249084,
|
| 8785 |
+
"learning_rate": 5.326278659611993e-05,
|
| 8786 |
+
"loss": 0.696,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 1.4711397597421623,
|
| 8791 |
+
"grad_norm": 0.39298853278160095,
|
| 8792 |
+
"learning_rate": 5.314520870076426e-05,
|
| 8793 |
+
"loss": 0.667,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 1.4723117491942572,
|
| 8798 |
+
"grad_norm": 0.3782164454460144,
|
| 8799 |
+
"learning_rate": 5.302763080540859e-05,
|
| 8800 |
+
"loss": 0.6258,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 1.4734837386463522,
|
| 8805 |
+
"grad_norm": 0.38333067297935486,
|
| 8806 |
+
"learning_rate": 5.291005291005291e-05,
|
| 8807 |
+
"loss": 0.666,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 1.4746557280984471,
|
| 8812 |
+
"grad_norm": 0.40639209747314453,
|
| 8813 |
+
"learning_rate": 5.279247501469724e-05,
|
| 8814 |
+
"loss": 0.6612,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 1.475827717550542,
|
| 8819 |
+
"grad_norm": 0.38118046522140503,
|
| 8820 |
+
"learning_rate": 5.267489711934157e-05,
|
| 8821 |
+
"loss": 0.6475,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 1.476999707002637,
|
| 8826 |
+
"grad_norm": 0.3647833466529846,
|
| 8827 |
+
"learning_rate": 5.255731922398589e-05,
|
| 8828 |
+
"loss": 0.6594,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 1.4781716964547318,
|
| 8833 |
+
"grad_norm": 0.3835331201553345,
|
| 8834 |
+
"learning_rate": 5.243974132863022e-05,
|
| 8835 |
+
"loss": 0.6208,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 1.479343685906827,
|
| 8840 |
+
"grad_norm": 0.3909193277359009,
|
| 8841 |
+
"learning_rate": 5.232216343327455e-05,
|
| 8842 |
+
"loss": 0.6997,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 1.4805156753589217,
|
| 8847 |
+
"grad_norm": 0.377839595079422,
|
| 8848 |
+
"learning_rate": 5.220458553791887e-05,
|
| 8849 |
+
"loss": 0.6687,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 1.4816876648110167,
|
| 8854 |
+
"grad_norm": 0.3836138844490051,
|
| 8855 |
+
"learning_rate": 5.20870076425632e-05,
|
| 8856 |
+
"loss": 0.6636,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 1.4828596542631116,
|
| 8861 |
+
"grad_norm": 0.37317249178886414,
|
| 8862 |
+
"learning_rate": 5.196942974720753e-05,
|
| 8863 |
+
"loss": 0.6793,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 1.4840316437152066,
|
| 8868 |
+
"grad_norm": 0.385690838098526,
|
| 8869 |
+
"learning_rate": 5.185185185185185e-05,
|
| 8870 |
+
"loss": 0.6517,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 1.4852036331673015,
|
| 8875 |
+
"grad_norm": 0.37199866771698,
|
| 8876 |
+
"learning_rate": 5.1734273956496184e-05,
|
| 8877 |
+
"loss": 0.6403,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 1.4863756226193965,
|
| 8882 |
+
"grad_norm": 0.393108606338501,
|
| 8883 |
+
"learning_rate": 5.161669606114051e-05,
|
| 8884 |
+
"loss": 0.7115,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 1.4875476120714914,
|
| 8889 |
+
"grad_norm": 0.40995368361473083,
|
| 8890 |
+
"learning_rate": 5.149911816578483e-05,
|
| 8891 |
+
"loss": 0.6859,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 1.4887196015235862,
|
| 8896 |
+
"grad_norm": 0.35813668370246887,
|
| 8897 |
+
"learning_rate": 5.1381540270429165e-05,
|
| 8898 |
+
"loss": 0.6265,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 1.4898915909756814,
|
| 8903 |
+
"grad_norm": 0.3874257504940033,
|
| 8904 |
+
"learning_rate": 5.126396237507349e-05,
|
| 8905 |
+
"loss": 0.6407,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 1.491063580427776,
|
| 8910 |
+
"grad_norm": 0.39107558131217957,
|
| 8911 |
+
"learning_rate": 5.114638447971781e-05,
|
| 8912 |
+
"loss": 0.6787,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 1.492235569879871,
|
| 8917 |
+
"grad_norm": 0.38405880331993103,
|
| 8918 |
+
"learning_rate": 5.102880658436214e-05,
|
| 8919 |
+
"loss": 0.6422,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 1.493407559331966,
|
| 8924 |
+
"grad_norm": 0.40102508664131165,
|
| 8925 |
+
"learning_rate": 5.0911228689006474e-05,
|
| 8926 |
+
"loss": 0.6573,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 1.494579548784061,
|
| 8931 |
+
"grad_norm": 0.4192945659160614,
|
| 8932 |
+
"learning_rate": 5.0793650793650794e-05,
|
| 8933 |
+
"loss": 0.6181,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 1.495751538236156,
|
| 8938 |
+
"grad_norm": 0.3956759572029114,
|
| 8939 |
+
"learning_rate": 5.067607289829512e-05,
|
| 8940 |
+
"loss": 0.6583,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 1.4969235276882509,
|
| 8945 |
+
"grad_norm": 0.4155990183353424,
|
| 8946 |
+
"learning_rate": 5.0558495002939455e-05,
|
| 8947 |
+
"loss": 0.6849,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 1.4980955171403458,
|
| 8952 |
+
"grad_norm": 0.390337198972702,
|
| 8953 |
+
"learning_rate": 5.0440917107583776e-05,
|
| 8954 |
+
"loss": 0.7172,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 1.4992675065924406,
|
| 8959 |
+
"grad_norm": 0.39508169889450073,
|
| 8960 |
+
"learning_rate": 5.03233392122281e-05,
|
| 8961 |
+
"loss": 0.6638,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 1.5004394960445357,
|
| 8966 |
+
"grad_norm": 0.3898437023162842,
|
| 8967 |
+
"learning_rate": 5.020576131687244e-05,
|
| 8968 |
+
"loss": 0.6085,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 1.5016114854966305,
|
| 8973 |
+
"grad_norm": 0.37632593512535095,
|
| 8974 |
+
"learning_rate": 5.008818342151676e-05,
|
| 8975 |
+
"loss": 0.6722,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 1.5027834749487254,
|
| 8980 |
+
"grad_norm": 0.39309123158454895,
|
| 8981 |
+
"learning_rate": 4.9970605526161084e-05,
|
| 8982 |
+
"loss": 0.7213,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 1.5039554644008204,
|
| 8987 |
+
"grad_norm": 0.3973393440246582,
|
| 8988 |
+
"learning_rate": 4.985302763080541e-05,
|
| 8989 |
+
"loss": 0.6851,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 1.5051274538529154,
|
| 8994 |
+
"grad_norm": 0.3938598930835724,
|
| 8995 |
+
"learning_rate": 4.973544973544973e-05,
|
| 8996 |
+
"loss": 0.6219,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 1.5062994433050103,
|
| 9001 |
+
"grad_norm": 0.38015884160995483,
|
| 9002 |
+
"learning_rate": 4.9617871840094066e-05,
|
| 9003 |
+
"loss": 0.7159,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 1.507471432757105,
|
| 9008 |
+
"grad_norm": 0.3672783076763153,
|
| 9009 |
+
"learning_rate": 4.950029394473839e-05,
|
| 9010 |
+
"loss": 0.6958,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 1.5086434222092002,
|
| 9015 |
+
"grad_norm": 0.3825698494911194,
|
| 9016 |
+
"learning_rate": 4.938271604938271e-05,
|
| 9017 |
+
"loss": 0.6572,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 1.509815411661295,
|
| 9022 |
+
"grad_norm": 0.38162460923194885,
|
| 9023 |
+
"learning_rate": 4.926513815402705e-05,
|
| 9024 |
+
"loss": 0.6262,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 1.5109874011133901,
|
| 9029 |
+
"grad_norm": 0.3847755789756775,
|
| 9030 |
+
"learning_rate": 4.9147560258671374e-05,
|
| 9031 |
+
"loss": 0.6155,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 1.5121593905654849,
|
| 9036 |
+
"grad_norm": 0.3806037902832031,
|
| 9037 |
+
"learning_rate": 4.9029982363315695e-05,
|
| 9038 |
+
"loss": 0.6606,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 1.5133313800175798,
|
| 9043 |
+
"grad_norm": 0.38434869050979614,
|
| 9044 |
+
"learning_rate": 4.891240446796003e-05,
|
| 9045 |
+
"loss": 0.687,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 1.5145033694696748,
|
| 9050 |
+
"grad_norm": 0.39996469020843506,
|
| 9051 |
+
"learning_rate": 4.879482657260435e-05,
|
| 9052 |
+
"loss": 0.6714,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 1.5156753589217697,
|
| 9057 |
+
"grad_norm": 0.37494775652885437,
|
| 9058 |
+
"learning_rate": 4.8677248677248676e-05,
|
| 9059 |
+
"loss": 0.669,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 1.5168473483738647,
|
| 9064 |
+
"grad_norm": 0.3752374053001404,
|
| 9065 |
+
"learning_rate": 4.855967078189301e-05,
|
| 9066 |
+
"loss": 0.6388,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 1.5180193378259594,
|
| 9071 |
+
"grad_norm": 0.36477306485176086,
|
| 9072 |
+
"learning_rate": 4.844209288653733e-05,
|
| 9073 |
+
"loss": 0.6531,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 1.5191913272780546,
|
| 9078 |
+
"grad_norm": 0.36245569586753845,
|
| 9079 |
+
"learning_rate": 4.832451499118166e-05,
|
| 9080 |
+
"loss": 0.6685,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 1.5203633167301494,
|
| 9085 |
+
"grad_norm": 0.355896919965744,
|
| 9086 |
+
"learning_rate": 4.820693709582599e-05,
|
| 9087 |
+
"loss": 0.6594,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 1.5215353061822443,
|
| 9092 |
+
"grad_norm": 0.3616688549518585,
|
| 9093 |
+
"learning_rate": 4.808935920047031e-05,
|
| 9094 |
+
"loss": 0.6425,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 1.5227072956343393,
|
| 9099 |
+
"grad_norm": 0.3755030632019043,
|
| 9100 |
+
"learning_rate": 4.797178130511464e-05,
|
| 9101 |
+
"loss": 0.6472,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 1.5238792850864342,
|
| 9106 |
+
"grad_norm": 0.3720192015171051,
|
| 9107 |
+
"learning_rate": 4.7854203409758966e-05,
|
| 9108 |
+
"loss": 0.6426,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 1.5250512745385292,
|
| 9113 |
+
"grad_norm": 0.3859194219112396,
|
| 9114 |
+
"learning_rate": 4.773662551440329e-05,
|
| 9115 |
+
"loss": 0.6385,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 1.5262232639906241,
|
| 9120 |
+
"grad_norm": 0.37670567631721497,
|
| 9121 |
+
"learning_rate": 4.761904761904762e-05,
|
| 9122 |
+
"loss": 0.6411,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 1.527395253442719,
|
| 9127 |
+
"grad_norm": 0.3701159358024597,
|
| 9128 |
+
"learning_rate": 4.750146972369195e-05,
|
| 9129 |
+
"loss": 0.6767,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 1.5285672428948138,
|
| 9134 |
+
"grad_norm": 0.38531991839408875,
|
| 9135 |
+
"learning_rate": 4.7383891828336275e-05,
|
| 9136 |
+
"loss": 0.6948,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 1.529739232346909,
|
| 9141 |
+
"grad_norm": 0.4071838855743408,
|
| 9142 |
+
"learning_rate": 4.72663139329806e-05,
|
| 9143 |
+
"loss": 0.6502,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 1.5309112217990037,
|
| 9148 |
+
"grad_norm": 0.38243693113327026,
|
| 9149 |
+
"learning_rate": 4.714873603762493e-05,
|
| 9150 |
+
"loss": 0.6327,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 1.5320832112510987,
|
| 9155 |
+
"grad_norm": 0.37889015674591064,
|
| 9156 |
+
"learning_rate": 4.7031158142269256e-05,
|
| 9157 |
+
"loss": 0.6182,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 1.5332552007031937,
|
| 9162 |
+
"grad_norm": 0.36006277799606323,
|
| 9163 |
+
"learning_rate": 4.691358024691358e-05,
|
| 9164 |
+
"loss": 0.6053,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 1.5344271901552886,
|
| 9169 |
+
"grad_norm": 0.39384907484054565,
|
| 9170 |
+
"learning_rate": 4.679600235155791e-05,
|
| 9171 |
+
"loss": 0.6609,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 1.5355991796073836,
|
| 9176 |
+
"grad_norm": 0.38327935338020325,
|
| 9177 |
+
"learning_rate": 4.667842445620224e-05,
|
| 9178 |
+
"loss": 0.6948,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 1.5367711690594783,
|
| 9183 |
+
"grad_norm": 0.38430511951446533,
|
| 9184 |
+
"learning_rate": 4.656084656084656e-05,
|
| 9185 |
+
"loss": 0.6313,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 1.5379431585115735,
|
| 9190 |
+
"grad_norm": 0.3972114324569702,
|
| 9191 |
+
"learning_rate": 4.644326866549089e-05,
|
| 9192 |
+
"loss": 0.6396,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 1.5391151479636682,
|
| 9197 |
+
"grad_norm": 0.4078161120414734,
|
| 9198 |
+
"learning_rate": 4.632569077013522e-05,
|
| 9199 |
+
"loss": 0.6885,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 1.5402871374157634,
|
| 9204 |
+
"grad_norm": 0.3729688227176666,
|
| 9205 |
+
"learning_rate": 4.620811287477954e-05,
|
| 9206 |
+
"loss": 0.6438,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 1.5414591268678581,
|
| 9211 |
+
"grad_norm": 0.3863990902900696,
|
| 9212 |
+
"learning_rate": 4.609053497942387e-05,
|
| 9213 |
+
"loss": 0.6424,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 1.542631116319953,
|
| 9218 |
+
"grad_norm": 0.3734455406665802,
|
| 9219 |
+
"learning_rate": 4.5972957084068194e-05,
|
| 9220 |
+
"loss": 0.6927,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 1.543803105772048,
|
| 9225 |
+
"grad_norm": 0.3718372583389282,
|
| 9226 |
+
"learning_rate": 4.585537918871252e-05,
|
| 9227 |
+
"loss": 0.6325,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 1.544975095224143,
|
| 9232 |
+
"grad_norm": 0.36497369408607483,
|
| 9233 |
+
"learning_rate": 4.5737801293356855e-05,
|
| 9234 |
+
"loss": 0.6187,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 1.546147084676238,
|
| 9239 |
+
"grad_norm": 0.38579320907592773,
|
| 9240 |
+
"learning_rate": 4.5620223398001175e-05,
|
| 9241 |
+
"loss": 0.6754,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 1.5473190741283327,
|
| 9246 |
+
"grad_norm": 0.374126136302948,
|
| 9247 |
+
"learning_rate": 4.55026455026455e-05,
|
| 9248 |
+
"loss": 0.6472,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 1.5484910635804279,
|
| 9253 |
+
"grad_norm": 0.38020315766334534,
|
| 9254 |
+
"learning_rate": 4.5385067607289836e-05,
|
| 9255 |
+
"loss": 0.672,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 1.5496630530325226,
|
| 9260 |
+
"grad_norm": 0.38782334327697754,
|
| 9261 |
+
"learning_rate": 4.5267489711934157e-05,
|
| 9262 |
+
"loss": 0.7069,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 1.5508350424846178,
|
| 9267 |
+
"grad_norm": 0.36838847398757935,
|
| 9268 |
+
"learning_rate": 4.5149911816578484e-05,
|
| 9269 |
+
"loss": 0.6567,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 1.5520070319367125,
|
| 9274 |
+
"grad_norm": 0.37560224533081055,
|
| 9275 |
+
"learning_rate": 4.503233392122281e-05,
|
| 9276 |
+
"loss": 0.6638,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 1.5531790213888075,
|
| 9281 |
+
"grad_norm": 0.3812626898288727,
|
| 9282 |
+
"learning_rate": 4.491475602586714e-05,
|
| 9283 |
+
"loss": 0.6841,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 1.5543510108409024,
|
| 9288 |
+
"grad_norm": 0.37894850969314575,
|
| 9289 |
+
"learning_rate": 4.4797178130511465e-05,
|
| 9290 |
+
"loss": 0.6505,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 1.5555230002929974,
|
| 9295 |
+
"grad_norm": 0.3598770499229431,
|
| 9296 |
+
"learning_rate": 4.467960023515579e-05,
|
| 9297 |
+
"loss": 0.651,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 1.5566949897450924,
|
| 9302 |
+
"grad_norm": 0.4064580798149109,
|
| 9303 |
+
"learning_rate": 4.456202233980012e-05,
|
| 9304 |
+
"loss": 0.6593,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 1.557866979197187,
|
| 9309 |
+
"grad_norm": 0.3867085576057434,
|
| 9310 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 9311 |
+
"loss": 0.6189,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 1.5590389686492823,
|
| 9316 |
+
"grad_norm": 0.3712484836578369,
|
| 9317 |
+
"learning_rate": 4.4326866549088774e-05,
|
| 9318 |
+
"loss": 0.6563,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 1.560210958101377,
|
| 9323 |
+
"grad_norm": 0.3924928605556488,
|
| 9324 |
+
"learning_rate": 4.42092886537331e-05,
|
| 9325 |
+
"loss": 0.6168,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 1.5613829475534722,
|
| 9330 |
+
"grad_norm": 0.36495327949523926,
|
| 9331 |
+
"learning_rate": 4.409171075837743e-05,
|
| 9332 |
+
"loss": 0.6584,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 1.562554937005567,
|
| 9337 |
+
"grad_norm": 0.37702497839927673,
|
| 9338 |
+
"learning_rate": 4.3974132863021755e-05,
|
| 9339 |
+
"loss": 0.6638,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 1.5637269264576619,
|
| 9344 |
+
"grad_norm": 0.3681667149066925,
|
| 9345 |
+
"learning_rate": 4.385655496766608e-05,
|
| 9346 |
+
"loss": 0.6122,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 1.5648989159097568,
|
| 9351 |
+
"grad_norm": 0.37110069394111633,
|
| 9352 |
+
"learning_rate": 4.37389770723104e-05,
|
| 9353 |
+
"loss": 0.6791,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 1.5660709053618518,
|
| 9358 |
+
"grad_norm": 0.3588317036628723,
|
| 9359 |
+
"learning_rate": 4.3621399176954737e-05,
|
| 9360 |
+
"loss": 0.616,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 1.5672428948139467,
|
| 9365 |
+
"grad_norm": 0.39171895384788513,
|
| 9366 |
+
"learning_rate": 4.3503821281599064e-05,
|
| 9367 |
+
"loss": 0.6859,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 1.5684148842660415,
|
| 9372 |
+
"grad_norm": 0.39346379041671753,
|
| 9373 |
+
"learning_rate": 4.3386243386243384e-05,
|
| 9374 |
+
"loss": 0.6833,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 1.5695868737181367,
|
| 9379 |
+
"grad_norm": 0.37361201643943787,
|
| 9380 |
+
"learning_rate": 4.326866549088772e-05,
|
| 9381 |
+
"loss": 0.637,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 1.5707588631702314,
|
| 9386 |
+
"grad_norm": 0.38074547052383423,
|
| 9387 |
+
"learning_rate": 4.315108759553204e-05,
|
| 9388 |
+
"loss": 0.6672,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 1.5719308526223263,
|
| 9393 |
+
"grad_norm": 0.36969512701034546,
|
| 9394 |
+
"learning_rate": 4.3033509700176366e-05,
|
| 9395 |
+
"loss": 0.6363,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 1.5731028420744213,
|
| 9400 |
+
"grad_norm": 0.36250782012939453,
|
| 9401 |
+
"learning_rate": 4.29159318048207e-05,
|
| 9402 |
+
"loss": 0.6194,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 1.5742748315265163,
|
| 9407 |
+
"grad_norm": 0.3632318377494812,
|
| 9408 |
+
"learning_rate": 4.279835390946502e-05,
|
| 9409 |
+
"loss": 0.6229,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 1.5754468209786112,
|
| 9414 |
+
"grad_norm": 0.37329503893852234,
|
| 9415 |
+
"learning_rate": 4.268077601410935e-05,
|
| 9416 |
+
"loss": 0.6614,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 1.5766188104307062,
|
| 9421 |
+
"grad_norm": 0.3797610402107239,
|
| 9422 |
+
"learning_rate": 4.256319811875368e-05,
|
| 9423 |
+
"loss": 0.6454,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 1.5777907998828011,
|
| 9428 |
+
"grad_norm": 0.38016003370285034,
|
| 9429 |
+
"learning_rate": 4.2445620223398e-05,
|
| 9430 |
+
"loss": 0.6362,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 1.5789627893348959,
|
| 9435 |
+
"grad_norm": 0.3744441568851471,
|
| 9436 |
+
"learning_rate": 4.232804232804233e-05,
|
| 9437 |
+
"loss": 0.6523,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 1.580134778786991,
|
| 9442 |
+
"grad_norm": 0.38999882340431213,
|
| 9443 |
+
"learning_rate": 4.2210464432686656e-05,
|
| 9444 |
+
"loss": 0.6723,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 1.5813067682390858,
|
| 9449 |
+
"grad_norm": 0.40173429250717163,
|
| 9450 |
+
"learning_rate": 4.209288653733098e-05,
|
| 9451 |
+
"loss": 0.6375,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 1.5824787576911807,
|
| 9456 |
+
"grad_norm": 0.38676032423973083,
|
| 9457 |
+
"learning_rate": 4.197530864197531e-05,
|
| 9458 |
+
"loss": 0.6853,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 1.5836507471432757,
|
| 9463 |
+
"grad_norm": 0.38119667768478394,
|
| 9464 |
+
"learning_rate": 4.185773074661964e-05,
|
| 9465 |
+
"loss": 0.6388,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 1.5848227365953707,
|
| 9470 |
+
"grad_norm": 0.3793523907661438,
|
| 9471 |
+
"learning_rate": 4.1740152851263964e-05,
|
| 9472 |
+
"loss": 0.6269,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 1.5859947260474656,
|
| 9477 |
+
"grad_norm": 0.3863115608692169,
|
| 9478 |
+
"learning_rate": 4.162257495590829e-05,
|
| 9479 |
+
"loss": 0.7104,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 1.5871667154995603,
|
| 9484 |
+
"grad_norm": 0.3677123785018921,
|
| 9485 |
+
"learning_rate": 4.150499706055262e-05,
|
| 9486 |
+
"loss": 0.6581,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 1.5883387049516555,
|
| 9491 |
+
"grad_norm": 0.39373883605003357,
|
| 9492 |
+
"learning_rate": 4.1387419165196946e-05,
|
| 9493 |
+
"loss": 0.6768,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 1.5895106944037503,
|
| 9498 |
+
"grad_norm": 0.39006727933883667,
|
| 9499 |
+
"learning_rate": 4.126984126984127e-05,
|
| 9500 |
+
"loss": 0.6636,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 1.5906826838558454,
|
| 9505 |
+
"grad_norm": 0.37299513816833496,
|
| 9506 |
+
"learning_rate": 4.11522633744856e-05,
|
| 9507 |
+
"loss": 0.6205,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 1.5918546733079402,
|
| 9512 |
+
"grad_norm": 0.374833345413208,
|
| 9513 |
+
"learning_rate": 4.103468547912993e-05,
|
| 9514 |
+
"loss": 0.6734,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 1.5930266627600351,
|
| 9519 |
+
"grad_norm": 0.38301679491996765,
|
| 9520 |
+
"learning_rate": 4.091710758377425e-05,
|
| 9521 |
+
"loss": 0.6709,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 1.59419865221213,
|
| 9526 |
+
"grad_norm": 0.35621005296707153,
|
| 9527 |
+
"learning_rate": 4.079952968841858e-05,
|
| 9528 |
+
"loss": 0.6523,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 1.595370641664225,
|
| 9533 |
+
"grad_norm": 0.3874707520008087,
|
| 9534 |
+
"learning_rate": 4.068195179306291e-05,
|
| 9535 |
+
"loss": 0.6615,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 1.59654263111632,
|
| 9540 |
+
"grad_norm": 0.39570900797843933,
|
| 9541 |
+
"learning_rate": 4.056437389770723e-05,
|
| 9542 |
+
"loss": 0.647,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 1.5977146205684147,
|
| 9547 |
+
"grad_norm": 0.38016220927238464,
|
| 9548 |
+
"learning_rate": 4.044679600235156e-05,
|
| 9549 |
+
"loss": 0.662,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 1.59888661002051,
|
| 9554 |
+
"grad_norm": 0.37574926018714905,
|
| 9555 |
+
"learning_rate": 4.032921810699588e-05,
|
| 9556 |
+
"loss": 0.6636,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 1.6000585994726046,
|
| 9561 |
+
"grad_norm": 0.40126630663871765,
|
| 9562 |
+
"learning_rate": 4.021164021164021e-05,
|
| 9563 |
+
"loss": 0.6081,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 1.6012305889246998,
|
| 9568 |
+
"grad_norm": 0.3685883581638336,
|
| 9569 |
+
"learning_rate": 4.0094062316284544e-05,
|
| 9570 |
+
"loss": 0.6364,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 1.6024025783767946,
|
| 9575 |
+
"grad_norm": 0.409110426902771,
|
| 9576 |
+
"learning_rate": 3.9976484420928865e-05,
|
| 9577 |
+
"loss": 0.7165,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 1.6035745678288895,
|
| 9582 |
+
"grad_norm": 0.39460107684135437,
|
| 9583 |
+
"learning_rate": 3.985890652557319e-05,
|
| 9584 |
+
"loss": 0.6756,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 1.6047465572809845,
|
| 9589 |
+
"grad_norm": 0.3784719407558441,
|
| 9590 |
+
"learning_rate": 3.9741328630217526e-05,
|
| 9591 |
+
"loss": 0.6578,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 1.6059185467330794,
|
| 9596 |
+
"grad_norm": 0.36832788586616516,
|
| 9597 |
+
"learning_rate": 3.9623750734861846e-05,
|
| 9598 |
+
"loss": 0.6372,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 1.6070905361851744,
|
| 9603 |
+
"grad_norm": 0.3807730972766876,
|
| 9604 |
+
"learning_rate": 3.950617283950617e-05,
|
| 9605 |
+
"loss": 0.6015,
|
| 9606 |
+
"step": 1371
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 1.6082625256372691,
|
| 9610 |
+
"grad_norm": 0.3825012445449829,
|
| 9611 |
+
"learning_rate": 3.93885949441505e-05,
|
| 9612 |
+
"loss": 0.6982,
|
| 9613 |
+
"step": 1372
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 1.6094345150893643,
|
| 9617 |
+
"grad_norm": 0.37073245644569397,
|
| 9618 |
+
"learning_rate": 3.927101704879483e-05,
|
| 9619 |
+
"loss": 0.6201,
|
| 9620 |
+
"step": 1373
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 1.610606504541459,
|
| 9624 |
+
"grad_norm": 0.38484933972358704,
|
| 9625 |
+
"learning_rate": 3.9153439153439155e-05,
|
| 9626 |
+
"loss": 0.6671,
|
| 9627 |
+
"step": 1374
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 1.6117784939935542,
|
| 9631 |
+
"grad_norm": 0.38665077090263367,
|
| 9632 |
+
"learning_rate": 3.903586125808348e-05,
|
| 9633 |
+
"loss": 0.6841,
|
| 9634 |
+
"step": 1375
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 1.612950483445649,
|
| 9638 |
+
"grad_norm": 0.3957137167453766,
|
| 9639 |
+
"learning_rate": 3.891828336272781e-05,
|
| 9640 |
+
"loss": 0.6734,
|
| 9641 |
+
"step": 1376
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 1.614122472897744,
|
| 9645 |
+
"grad_norm": 0.3841049075126648,
|
| 9646 |
+
"learning_rate": 3.8800705467372136e-05,
|
| 9647 |
+
"loss": 0.642,
|
| 9648 |
+
"step": 1377
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 1.6152944623498389,
|
| 9652 |
+
"grad_norm": 0.3854089081287384,
|
| 9653 |
+
"learning_rate": 3.868312757201646e-05,
|
| 9654 |
+
"loss": 0.6292,
|
| 9655 |
+
"step": 1378
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 1.6164664518019338,
|
| 9659 |
+
"grad_norm": 0.37589502334594727,
|
| 9660 |
+
"learning_rate": 3.856554967666079e-05,
|
| 9661 |
+
"loss": 0.6337,
|
| 9662 |
+
"step": 1379
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 1.6176384412540288,
|
| 9666 |
+
"grad_norm": 0.3719007968902588,
|
| 9667 |
+
"learning_rate": 3.844797178130512e-05,
|
| 9668 |
+
"loss": 0.6396,
|
| 9669 |
+
"step": 1380
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 1.6188104307061235,
|
| 9673 |
+
"grad_norm": 0.4034193158149719,
|
| 9674 |
+
"learning_rate": 3.8330393885949445e-05,
|
| 9675 |
+
"loss": 0.6406,
|
| 9676 |
+
"step": 1381
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 1.6199824201582187,
|
| 9680 |
+
"grad_norm": 0.39838656783103943,
|
| 9681 |
+
"learning_rate": 3.821281599059377e-05,
|
| 9682 |
+
"loss": 0.674,
|
| 9683 |
+
"step": 1382
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 1.6211544096103134,
|
| 9687 |
+
"grad_norm": 0.3712717592716217,
|
| 9688 |
+
"learning_rate": 3.809523809523809e-05,
|
| 9689 |
+
"loss": 0.6493,
|
| 9690 |
+
"step": 1383
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 1.6223263990624084,
|
| 9694 |
+
"grad_norm": 0.38980063796043396,
|
| 9695 |
+
"learning_rate": 3.7977660199882426e-05,
|
| 9696 |
+
"loss": 0.6588,
|
| 9697 |
+
"step": 1384
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 1.6234983885145033,
|
| 9701 |
+
"grad_norm": 0.37699729204177856,
|
| 9702 |
+
"learning_rate": 3.786008230452675e-05,
|
| 9703 |
+
"loss": 0.6591,
|
| 9704 |
+
"step": 1385
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 1.6246703779665983,
|
| 9708 |
+
"grad_norm": 0.3779825270175934,
|
| 9709 |
+
"learning_rate": 3.7742504409171074e-05,
|
| 9710 |
+
"loss": 0.6293,
|
| 9711 |
+
"step": 1386
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 1.6258423674186933,
|
| 9715 |
+
"grad_norm": 0.40519246459007263,
|
| 9716 |
+
"learning_rate": 3.762492651381541e-05,
|
| 9717 |
+
"loss": 0.6035,
|
| 9718 |
+
"step": 1387
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 1.6270143568707882,
|
| 9722 |
+
"grad_norm": 0.3740384876728058,
|
| 9723 |
+
"learning_rate": 3.750734861845973e-05,
|
| 9724 |
+
"loss": 0.6495,
|
| 9725 |
+
"step": 1388
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 1.6281863463228832,
|
| 9729 |
+
"grad_norm": 0.393038272857666,
|
| 9730 |
+
"learning_rate": 3.7389770723104055e-05,
|
| 9731 |
+
"loss": 0.6375,
|
| 9732 |
+
"step": 1389
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 1.629358335774978,
|
| 9736 |
+
"grad_norm": 0.3672095835208893,
|
| 9737 |
+
"learning_rate": 3.727219282774839e-05,
|
| 9738 |
+
"loss": 0.6262,
|
| 9739 |
+
"step": 1390
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 1.630530325227073,
|
| 9743 |
+
"grad_norm": 0.36593756079673767,
|
| 9744 |
+
"learning_rate": 3.715461493239271e-05,
|
| 9745 |
+
"loss": 0.6107,
|
| 9746 |
+
"step": 1391
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 1.6317023146791678,
|
| 9750 |
+
"grad_norm": 0.38510987162590027,
|
| 9751 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 9752 |
+
"loss": 0.666,
|
| 9753 |
+
"step": 1392
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 1.6328743041312628,
|
| 9757 |
+
"grad_norm": 0.37760478258132935,
|
| 9758 |
+
"learning_rate": 3.691945914168137e-05,
|
| 9759 |
+
"loss": 0.6047,
|
| 9760 |
+
"step": 1393
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 1.6340462935833577,
|
| 9764 |
+
"grad_norm": 0.39161157608032227,
|
| 9765 |
+
"learning_rate": 3.680188124632569e-05,
|
| 9766 |
+
"loss": 0.6353,
|
| 9767 |
+
"step": 1394
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 1.6352182830354527,
|
| 9771 |
+
"grad_norm": 0.4048272669315338,
|
| 9772 |
+
"learning_rate": 3.668430335097002e-05,
|
| 9773 |
+
"loss": 0.6406,
|
| 9774 |
+
"step": 1395
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 1.6363902724875476,
|
| 9778 |
+
"grad_norm": 0.3909299671649933,
|
| 9779 |
+
"learning_rate": 3.6566725455614345e-05,
|
| 9780 |
+
"loss": 0.6602,
|
| 9781 |
+
"step": 1396
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 1.6375622619396424,
|
| 9785 |
+
"grad_norm": 0.3780452609062195,
|
| 9786 |
+
"learning_rate": 3.644914756025867e-05,
|
| 9787 |
+
"loss": 0.6563,
|
| 9788 |
+
"step": 1397
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 1.6387342513917376,
|
| 9792 |
+
"grad_norm": 0.415984183549881,
|
| 9793 |
+
"learning_rate": 3.6331569664903e-05,
|
| 9794 |
+
"loss": 0.6956,
|
| 9795 |
+
"step": 1398
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 1.6399062408438323,
|
| 9799 |
+
"grad_norm": 0.39058148860931396,
|
| 9800 |
+
"learning_rate": 3.6213991769547327e-05,
|
| 9801 |
+
"loss": 0.618,
|
| 9802 |
+
"step": 1399
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 1.6410782302959275,
|
| 9806 |
+
"grad_norm": 0.3820166289806366,
|
| 9807 |
+
"learning_rate": 3.6096413874191654e-05,
|
| 9808 |
+
"loss": 0.6871,
|
| 9809 |
+
"step": 1400
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 1.6422502197480222,
|
| 9813 |
+
"grad_norm": 0.3850885331630707,
|
| 9814 |
+
"learning_rate": 3.597883597883598e-05,
|
| 9815 |
+
"loss": 0.6564,
|
| 9816 |
+
"step": 1401
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 1.6434222092001172,
|
| 9820 |
+
"grad_norm": 0.36710235476493835,
|
| 9821 |
+
"learning_rate": 3.586125808348031e-05,
|
| 9822 |
+
"loss": 0.6465,
|
| 9823 |
+
"step": 1402
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 1.6445941986522121,
|
| 9827 |
+
"grad_norm": 0.37936750054359436,
|
| 9828 |
+
"learning_rate": 3.5743680188124635e-05,
|
| 9829 |
+
"loss": 0.6541,
|
| 9830 |
+
"step": 1403
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 1.645766188104307,
|
| 9834 |
+
"grad_norm": 0.4197543263435364,
|
| 9835 |
+
"learning_rate": 3.562610229276896e-05,
|
| 9836 |
+
"loss": 0.7147,
|
| 9837 |
+
"step": 1404
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 1.646938177556402,
|
| 9841 |
+
"grad_norm": 0.39201852679252625,
|
| 9842 |
+
"learning_rate": 3.550852439741329e-05,
|
| 9843 |
+
"loss": 0.6829,
|
| 9844 |
+
"step": 1405
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 1.6481101670084968,
|
| 9848 |
+
"grad_norm": 0.3781983554363251,
|
| 9849 |
+
"learning_rate": 3.539094650205762e-05,
|
| 9850 |
+
"loss": 0.6482,
|
| 9851 |
+
"step": 1406
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 1.649282156460592,
|
| 9855 |
+
"grad_norm": 0.3771427869796753,
|
| 9856 |
+
"learning_rate": 3.527336860670194e-05,
|
| 9857 |
+
"loss": 0.6526,
|
| 9858 |
+
"step": 1407
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 1.6504541459126867,
|
| 9862 |
+
"grad_norm": 0.38876059651374817,
|
| 9863 |
+
"learning_rate": 3.515579071134627e-05,
|
| 9864 |
+
"loss": 0.6731,
|
| 9865 |
+
"step": 1408
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 1.6516261353647819,
|
| 9869 |
+
"grad_norm": 0.38884153962135315,
|
| 9870 |
+
"learning_rate": 3.50382128159906e-05,
|
| 9871 |
+
"loss": 0.6311,
|
| 9872 |
+
"step": 1409
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 1.6527981248168766,
|
| 9876 |
+
"grad_norm": 0.39201685786247253,
|
| 9877 |
+
"learning_rate": 3.492063492063492e-05,
|
| 9878 |
+
"loss": 0.6902,
|
| 9879 |
+
"step": 1410
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 1.6539701142689716,
|
| 9883 |
+
"grad_norm": 0.37352827191352844,
|
| 9884 |
+
"learning_rate": 3.480305702527925e-05,
|
| 9885 |
+
"loss": 0.6938,
|
| 9886 |
+
"step": 1411
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 1.6551421037210665,
|
| 9890 |
+
"grad_norm": 0.37906309962272644,
|
| 9891 |
+
"learning_rate": 3.468547912992357e-05,
|
| 9892 |
+
"loss": 0.6818,
|
| 9893 |
+
"step": 1412
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 1.6563140931731615,
|
| 9897 |
+
"grad_norm": 0.3947876989841461,
|
| 9898 |
+
"learning_rate": 3.45679012345679e-05,
|
| 9899 |
+
"loss": 0.6529,
|
| 9900 |
+
"step": 1413
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 1.6574860826252564,
|
| 9904 |
+
"grad_norm": 0.3685528635978699,
|
| 9905 |
+
"learning_rate": 3.4450323339212234e-05,
|
| 9906 |
+
"loss": 0.6282,
|
| 9907 |
+
"step": 1414
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 1.6586580720773512,
|
| 9911 |
+
"grad_norm": 0.3796926438808441,
|
| 9912 |
+
"learning_rate": 3.4332745443856554e-05,
|
| 9913 |
+
"loss": 0.6472,
|
| 9914 |
+
"step": 1415
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 1.6598300615294463,
|
| 9918 |
+
"grad_norm": 0.3854847848415375,
|
| 9919 |
+
"learning_rate": 3.421516754850088e-05,
|
| 9920 |
+
"loss": 0.6531,
|
| 9921 |
+
"step": 1416
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 1.661002050981541,
|
| 9925 |
+
"grad_norm": 0.3648832440376282,
|
| 9926 |
+
"learning_rate": 3.4097589653145215e-05,
|
| 9927 |
+
"loss": 0.6203,
|
| 9928 |
+
"step": 1417
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 1.6621740404336363,
|
| 9932 |
+
"grad_norm": 0.37867122888565063,
|
| 9933 |
+
"learning_rate": 3.3980011757789536e-05,
|
| 9934 |
+
"loss": 0.657,
|
| 9935 |
+
"step": 1418
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 1.663346029885731,
|
| 9939 |
+
"grad_norm": 0.3739066421985626,
|
| 9940 |
+
"learning_rate": 3.386243386243386e-05,
|
| 9941 |
+
"loss": 0.6555,
|
| 9942 |
+
"step": 1419
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 1.664518019337826,
|
| 9946 |
+
"grad_norm": 0.3743496537208557,
|
| 9947 |
+
"learning_rate": 3.374485596707819e-05,
|
| 9948 |
+
"loss": 0.6406,
|
| 9949 |
+
"step": 1420
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 1.665690008789921,
|
| 9953 |
+
"grad_norm": 0.38990095257759094,
|
| 9954 |
+
"learning_rate": 3.362727807172252e-05,
|
| 9955 |
+
"loss": 0.6006,
|
| 9956 |
+
"step": 1421
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 1.6668619982420159,
|
| 9960 |
+
"grad_norm": 0.38039836287498474,
|
| 9961 |
+
"learning_rate": 3.3509700176366844e-05,
|
| 9962 |
+
"loss": 0.6363,
|
| 9963 |
+
"step": 1422
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 1.6680339876941108,
|
| 9967 |
+
"grad_norm": 0.36417829990386963,
|
| 9968 |
+
"learning_rate": 3.339212228101117e-05,
|
| 9969 |
+
"loss": 0.6366,
|
| 9970 |
+
"step": 1423
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 1.6692059771462056,
|
| 9974 |
+
"grad_norm": 0.37726080417633057,
|
| 9975 |
+
"learning_rate": 3.32745443856555e-05,
|
| 9976 |
+
"loss": 0.6574,
|
| 9977 |
+
"step": 1424
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 1.6703779665983007,
|
| 9981 |
+
"grad_norm": 0.3663460910320282,
|
| 9982 |
+
"learning_rate": 3.3156966490299826e-05,
|
| 9983 |
+
"loss": 0.6306,
|
| 9984 |
+
"step": 1425
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 1.6715499560503955,
|
| 9988 |
+
"grad_norm": 0.4453365206718445,
|
| 9989 |
+
"learning_rate": 3.303938859494415e-05,
|
| 9990 |
+
"loss": 0.6846,
|
| 9991 |
+
"step": 1426
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 1.6727219455024904,
|
| 9995 |
+
"grad_norm": 0.3600301742553711,
|
| 9996 |
+
"learning_rate": 3.292181069958848e-05,
|
| 9997 |
+
"loss": 0.6361,
|
| 9998 |
+
"step": 1427
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 1.6738939349545854,
|
| 10002 |
+
"grad_norm": 0.3683728575706482,
|
| 10003 |
+
"learning_rate": 3.280423280423281e-05,
|
| 10004 |
+
"loss": 0.6304,
|
| 10005 |
+
"step": 1428
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 1.6750659244066803,
|
| 10009 |
+
"grad_norm": 0.4204603135585785,
|
| 10010 |
+
"learning_rate": 3.2686654908877134e-05,
|
| 10011 |
+
"loss": 0.6582,
|
| 10012 |
+
"step": 1429
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 1.6762379138587753,
|
| 10016 |
+
"grad_norm": 0.3554556667804718,
|
| 10017 |
+
"learning_rate": 3.256907701352146e-05,
|
| 10018 |
+
"loss": 0.6506,
|
| 10019 |
+
"step": 1430
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 1.6774099033108703,
|
| 10023 |
+
"grad_norm": 0.39126551151275635,
|
| 10024 |
+
"learning_rate": 3.245149911816578e-05,
|
| 10025 |
+
"loss": 0.6441,
|
| 10026 |
+
"step": 1431
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 1.6785818927629652,
|
| 10030 |
+
"grad_norm": 0.37401652336120605,
|
| 10031 |
+
"learning_rate": 3.2333921222810116e-05,
|
| 10032 |
+
"loss": 0.6569,
|
| 10033 |
+
"step": 1432
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 1.67975388221506,
|
| 10037 |
+
"grad_norm": 0.38046538829803467,
|
| 10038 |
+
"learning_rate": 3.221634332745444e-05,
|
| 10039 |
+
"loss": 0.642,
|
| 10040 |
+
"step": 1433
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 1.6809258716671551,
|
| 10044 |
+
"grad_norm": 0.375357061624527,
|
| 10045 |
+
"learning_rate": 3.209876543209876e-05,
|
| 10046 |
+
"loss": 0.66,
|
| 10047 |
+
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 1.6820978611192499,
|
| 10051 |
+
"grad_norm": 0.36403539776802063,
|
| 10052 |
+
"learning_rate": 3.19811875367431e-05,
|
| 10053 |
+
"loss": 0.6493,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 1.6832698505713448,
|
| 10058 |
+
"grad_norm": 0.3747311234474182,
|
| 10059 |
+
"learning_rate": 3.186360964138742e-05,
|
| 10060 |
+
"loss": 0.6422,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 1.6844418400234398,
|
| 10065 |
+
"grad_norm": 0.38533565402030945,
|
| 10066 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 10067 |
+
"loss": 0.6381,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 1.6856138294755347,
|
| 10072 |
+
"grad_norm": 0.38410523533821106,
|
| 10073 |
+
"learning_rate": 3.162845385067608e-05,
|
| 10074 |
+
"loss": 0.6695,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 1.6867858189276297,
|
| 10079 |
+
"grad_norm": 0.37452232837677,
|
| 10080 |
+
"learning_rate": 3.15108759553204e-05,
|
| 10081 |
+
"loss": 0.6305,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 1.6879578083797244,
|
| 10086 |
+
"grad_norm": 0.3752080500125885,
|
| 10087 |
+
"learning_rate": 3.1393298059964726e-05,
|
| 10088 |
+
"loss": 0.6486,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 1.6891297978318196,
|
| 10093 |
+
"grad_norm": 0.384127676486969,
|
| 10094 |
+
"learning_rate": 3.127572016460906e-05,
|
| 10095 |
+
"loss": 0.628,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 1.6903017872839143,
|
| 10100 |
+
"grad_norm": 0.3655785024166107,
|
| 10101 |
+
"learning_rate": 3.115814226925338e-05,
|
| 10102 |
+
"loss": 0.6325,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 1.6914737767360095,
|
| 10107 |
+
"grad_norm": 0.37582463026046753,
|
| 10108 |
+
"learning_rate": 3.104056437389771e-05,
|
| 10109 |
+
"loss": 0.6641,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 1.6926457661881043,
|
| 10114 |
+
"grad_norm": 0.3942627012729645,
|
| 10115 |
+
"learning_rate": 3.0922986478542035e-05,
|
| 10116 |
+
"loss": 0.6314,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 1.6938177556401992,
|
| 10121 |
+
"grad_norm": 0.38446322083473206,
|
| 10122 |
+
"learning_rate": 3.080540858318636e-05,
|
| 10123 |
+
"loss": 0.6482,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 1.6949897450922942,
|
| 10128 |
+
"grad_norm": 0.3782792091369629,
|
| 10129 |
+
"learning_rate": 3.068783068783069e-05,
|
| 10130 |
+
"loss": 0.666,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 1.6961617345443891,
|
| 10135 |
+
"grad_norm": 0.3792760372161865,
|
| 10136 |
+
"learning_rate": 3.0570252792475016e-05,
|
| 10137 |
+
"loss": 0.6424,
|
| 10138 |
+
"step": 1447
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 1.697333723996484,
|
| 10142 |
+
"grad_norm": 0.4009881615638733,
|
| 10143 |
+
"learning_rate": 3.0452674897119343e-05,
|
| 10144 |
+
"loss": 0.6803,
|
| 10145 |
+
"step": 1448
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 1.6985057134485788,
|
| 10149 |
+
"grad_norm": 0.3797784149646759,
|
| 10150 |
+
"learning_rate": 3.0335097001763667e-05,
|
| 10151 |
+
"loss": 0.6243,
|
| 10152 |
+
"step": 1449
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 1.699677702900674,
|
| 10156 |
+
"grad_norm": 0.37865757942199707,
|
| 10157 |
+
"learning_rate": 3.0217519106407998e-05,
|
| 10158 |
+
"loss": 0.6354,
|
| 10159 |
+
"step": 1450
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 1.7008496923527687,
|
| 10163 |
+
"grad_norm": 0.38236260414123535,
|
| 10164 |
+
"learning_rate": 3.0099941211052325e-05,
|
| 10165 |
+
"loss": 0.6653,
|
| 10166 |
+
"step": 1451
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 1.702021681804864,
|
| 10170 |
+
"grad_norm": 0.37716102600097656,
|
| 10171 |
+
"learning_rate": 2.998236331569665e-05,
|
| 10172 |
+
"loss": 0.6721,
|
| 10173 |
+
"step": 1452
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 1.7031936712569586,
|
| 10177 |
+
"grad_norm": 0.38444986939430237,
|
| 10178 |
+
"learning_rate": 2.986478542034098e-05,
|
| 10179 |
+
"loss": 0.6269,
|
| 10180 |
+
"step": 1453
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 1.7043656607090536,
|
| 10184 |
+
"grad_norm": 0.3900532126426697,
|
| 10185 |
+
"learning_rate": 2.9747207524985303e-05,
|
| 10186 |
+
"loss": 0.6657,
|
| 10187 |
+
"step": 1454
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 1.7055376501611486,
|
| 10191 |
+
"grad_norm": 0.3843390643596649,
|
| 10192 |
+
"learning_rate": 2.962962962962963e-05,
|
| 10193 |
+
"loss": 0.6847,
|
| 10194 |
+
"step": 1455
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 1.7067096396132435,
|
| 10198 |
+
"grad_norm": 0.3660966753959656,
|
| 10199 |
+
"learning_rate": 2.951205173427396e-05,
|
| 10200 |
+
"loss": 0.6667,
|
| 10201 |
+
"step": 1456
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 1.7078816290653385,
|
| 10205 |
+
"grad_norm": 0.3676696717739105,
|
| 10206 |
+
"learning_rate": 2.9394473838918284e-05,
|
| 10207 |
+
"loss": 0.6643,
|
| 10208 |
+
"step": 1457
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 1.7090536185174332,
|
| 10212 |
+
"grad_norm": 0.3827989101409912,
|
| 10213 |
+
"learning_rate": 2.927689594356261e-05,
|
| 10214 |
+
"loss": 0.6637,
|
| 10215 |
+
"step": 1458
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 1.7102256079695284,
|
| 10219 |
+
"grad_norm": 0.3773452639579773,
|
| 10220 |
+
"learning_rate": 2.9159318048206942e-05,
|
| 10221 |
+
"loss": 0.6432,
|
| 10222 |
+
"step": 1459
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 1.7113975974216231,
|
| 10226 |
+
"grad_norm": 0.39179813861846924,
|
| 10227 |
+
"learning_rate": 2.9041740152851266e-05,
|
| 10228 |
+
"loss": 0.6571,
|
| 10229 |
+
"step": 1460
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 1.7125695868737183,
|
| 10233 |
+
"grad_norm": 0.3847814202308655,
|
| 10234 |
+
"learning_rate": 2.892416225749559e-05,
|
| 10235 |
+
"loss": 0.6611,
|
| 10236 |
+
"step": 1461
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 1.713741576325813,
|
| 10240 |
+
"grad_norm": 0.3790866732597351,
|
| 10241 |
+
"learning_rate": 2.880658436213992e-05,
|
| 10242 |
+
"loss": 0.6731,
|
| 10243 |
+
"step": 1462
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 1.714913565777908,
|
| 10247 |
+
"grad_norm": 0.37719854712486267,
|
| 10248 |
+
"learning_rate": 2.8689006466784247e-05,
|
| 10249 |
+
"loss": 0.6456,
|
| 10250 |
+
"step": 1463
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 1.716085555230003,
|
| 10254 |
+
"grad_norm": 0.3641086220741272,
|
| 10255 |
+
"learning_rate": 2.857142857142857e-05,
|
| 10256 |
+
"loss": 0.6548,
|
| 10257 |
+
"step": 1464
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 1.717257544682098,
|
| 10261 |
+
"grad_norm": 0.39434361457824707,
|
| 10262 |
+
"learning_rate": 2.84538506760729e-05,
|
| 10263 |
+
"loss": 0.6535,
|
| 10264 |
+
"step": 1465
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 1.7184295341341929,
|
| 10268 |
+
"grad_norm": 0.36238378286361694,
|
| 10269 |
+
"learning_rate": 2.8336272780717225e-05,
|
| 10270 |
+
"loss": 0.6303,
|
| 10271 |
+
"step": 1466
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 1.7196015235862876,
|
| 10275 |
+
"grad_norm": 0.3763822615146637,
|
| 10276 |
+
"learning_rate": 2.8218694885361552e-05,
|
| 10277 |
+
"loss": 0.7062,
|
| 10278 |
+
"step": 1467
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 1.7207735130383828,
|
| 10282 |
+
"grad_norm": 0.37111690640449524,
|
| 10283 |
+
"learning_rate": 2.8101116990005883e-05,
|
| 10284 |
+
"loss": 0.6468,
|
| 10285 |
+
"step": 1468
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 1.7219455024904775,
|
| 10289 |
+
"grad_norm": 0.3747517764568329,
|
| 10290 |
+
"learning_rate": 2.7983539094650207e-05,
|
| 10291 |
+
"loss": 0.663,
|
| 10292 |
+
"step": 1469
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 1.7231174919425725,
|
| 10296 |
+
"grad_norm": 0.3822534382343292,
|
| 10297 |
+
"learning_rate": 2.7865961199294534e-05,
|
| 10298 |
+
"loss": 0.6306,
|
| 10299 |
+
"step": 1470
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 1.7242894813946674,
|
| 10303 |
+
"grad_norm": 0.36476072669029236,
|
| 10304 |
+
"learning_rate": 2.7748383303938864e-05,
|
| 10305 |
+
"loss": 0.6283,
|
| 10306 |
+
"step": 1471
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 1.7254614708467624,
|
| 10310 |
+
"grad_norm": 0.36173558235168457,
|
| 10311 |
+
"learning_rate": 2.7630805408583188e-05,
|
| 10312 |
+
"loss": 0.618,
|
| 10313 |
+
"step": 1472
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 1.7266334602988573,
|
| 10317 |
+
"grad_norm": 0.36746352910995483,
|
| 10318 |
+
"learning_rate": 2.7513227513227512e-05,
|
| 10319 |
+
"loss": 0.6457,
|
| 10320 |
+
"step": 1473
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 1.7278054497509523,
|
| 10324 |
+
"grad_norm": 0.37200871109962463,
|
| 10325 |
+
"learning_rate": 2.7395649617871842e-05,
|
| 10326 |
+
"loss": 0.6204,
|
| 10327 |
+
"step": 1474
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 1.7289774392030473,
|
| 10331 |
+
"grad_norm": 0.3685344159603119,
|
| 10332 |
+
"learning_rate": 2.727807172251617e-05,
|
| 10333 |
+
"loss": 0.6303,
|
| 10334 |
+
"step": 1475
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 1.730149428655142,
|
| 10338 |
+
"grad_norm": 0.37158438563346863,
|
| 10339 |
+
"learning_rate": 2.7160493827160493e-05,
|
| 10340 |
+
"loss": 0.6278,
|
| 10341 |
+
"step": 1476
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 1.7313214181072372,
|
| 10345 |
+
"grad_norm": 0.3743583559989929,
|
| 10346 |
+
"learning_rate": 2.7042915931804824e-05,
|
| 10347 |
+
"loss": 0.6088,
|
| 10348 |
+
"step": 1477
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 1.732493407559332,
|
| 10352 |
+
"grad_norm": 0.37238809466362,
|
| 10353 |
+
"learning_rate": 2.6925338036449148e-05,
|
| 10354 |
+
"loss": 0.6456,
|
| 10355 |
+
"step": 1478
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 1.7336653970114269,
|
| 10359 |
+
"grad_norm": 0.3718564808368683,
|
| 10360 |
+
"learning_rate": 2.6807760141093475e-05,
|
| 10361 |
+
"loss": 0.6123,
|
| 10362 |
+
"step": 1479
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 1.7348373864635218,
|
| 10366 |
+
"grad_norm": 0.3883103132247925,
|
| 10367 |
+
"learning_rate": 2.6690182245737805e-05,
|
| 10368 |
+
"loss": 0.6628,
|
| 10369 |
+
"step": 1480
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 1.7360093759156168,
|
| 10373 |
+
"grad_norm": 0.376810759305954,
|
| 10374 |
+
"learning_rate": 2.657260435038213e-05,
|
| 10375 |
+
"loss": 0.6465,
|
| 10376 |
+
"step": 1481
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 1.7371813653677117,
|
| 10380 |
+
"grad_norm": 0.41317370533943176,
|
| 10381 |
+
"learning_rate": 2.6455026455026456e-05,
|
| 10382 |
+
"loss": 0.6835,
|
| 10383 |
+
"step": 1482
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 1.7383533548198065,
|
| 10387 |
+
"grad_norm": 0.3698785603046417,
|
| 10388 |
+
"learning_rate": 2.6337448559670787e-05,
|
| 10389 |
+
"loss": 0.6174,
|
| 10390 |
+
"step": 1483
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 1.7395253442719016,
|
| 10394 |
+
"grad_norm": 0.3769229054450989,
|
| 10395 |
+
"learning_rate": 2.621987066431511e-05,
|
| 10396 |
+
"loss": 0.6402,
|
| 10397 |
+
"step": 1484
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 1.7406973337239964,
|
| 10401 |
+
"grad_norm": 0.3612512946128845,
|
| 10402 |
+
"learning_rate": 2.6102292768959434e-05,
|
| 10403 |
+
"loss": 0.6457,
|
| 10404 |
+
"step": 1485
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 1.7418693231760916,
|
| 10408 |
+
"grad_norm": 0.36678725481033325,
|
| 10409 |
+
"learning_rate": 2.5984714873603765e-05,
|
| 10410 |
+
"loss": 0.6247,
|
| 10411 |
+
"step": 1486
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 1.7430413126281863,
|
| 10415 |
+
"grad_norm": 0.39542651176452637,
|
| 10416 |
+
"learning_rate": 2.5867136978248092e-05,
|
| 10417 |
+
"loss": 0.6294,
|
| 10418 |
+
"step": 1487
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 1.7442133020802812,
|
| 10422 |
+
"grad_norm": 0.3900982439517975,
|
| 10423 |
+
"learning_rate": 2.5749559082892416e-05,
|
| 10424 |
+
"loss": 0.675,
|
| 10425 |
+
"step": 1488
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 1.7453852915323762,
|
| 10429 |
+
"grad_norm": 0.3719351887702942,
|
| 10430 |
+
"learning_rate": 2.5631981187536746e-05,
|
| 10431 |
+
"loss": 0.6492,
|
| 10432 |
+
"step": 1489
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 1.7465572809844712,
|
| 10436 |
+
"grad_norm": 0.37937313318252563,
|
| 10437 |
+
"learning_rate": 2.551440329218107e-05,
|
| 10438 |
+
"loss": 0.6232,
|
| 10439 |
+
"step": 1490
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 1.7477292704365661,
|
| 10443 |
+
"grad_norm": 0.39128735661506653,
|
| 10444 |
+
"learning_rate": 2.5396825396825397e-05,
|
| 10445 |
+
"loss": 0.6404,
|
| 10446 |
+
"step": 1491
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 1.7489012598886609,
|
| 10450 |
+
"grad_norm": 0.3798913359642029,
|
| 10451 |
+
"learning_rate": 2.5279247501469728e-05,
|
| 10452 |
+
"loss": 0.6588,
|
| 10453 |
+
"step": 1492
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 1.750073249340756,
|
| 10457 |
+
"grad_norm": 0.3671737015247345,
|
| 10458 |
+
"learning_rate": 2.516166960611405e-05,
|
| 10459 |
+
"loss": 0.671,
|
| 10460 |
+
"step": 1493
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 1.7512452387928508,
|
| 10464 |
+
"grad_norm": 0.3778979182243347,
|
| 10465 |
+
"learning_rate": 2.504409171075838e-05,
|
| 10466 |
+
"loss": 0.6157,
|
| 10467 |
+
"step": 1494
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 1.752417228244946,
|
| 10471 |
+
"grad_norm": 0.38095155358314514,
|
| 10472 |
+
"learning_rate": 2.4926513815402706e-05,
|
| 10473 |
+
"loss": 0.6674,
|
| 10474 |
+
"step": 1495
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 1.7535892176970407,
|
| 10478 |
+
"grad_norm": 0.39539647102355957,
|
| 10479 |
+
"learning_rate": 2.4808935920047033e-05,
|
| 10480 |
+
"loss": 0.6457,
|
| 10481 |
+
"step": 1496
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 1.7547612071491356,
|
| 10485 |
+
"grad_norm": 0.3683115541934967,
|
| 10486 |
+
"learning_rate": 2.4691358024691357e-05,
|
| 10487 |
+
"loss": 0.6207,
|
| 10488 |
+
"step": 1497
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 1.7559331966012306,
|
| 10492 |
+
"grad_norm": 0.38240090012550354,
|
| 10493 |
+
"learning_rate": 2.4573780129335687e-05,
|
| 10494 |
+
"loss": 0.6593,
|
| 10495 |
+
"step": 1498
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 1.7571051860533256,
|
| 10499 |
+
"grad_norm": 0.36555352807044983,
|
| 10500 |
+
"learning_rate": 2.4456202233980014e-05,
|
| 10501 |
+
"loss": 0.6818,
|
| 10502 |
+
"step": 1499
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 1.7582771755054205,
|
| 10506 |
+
"grad_norm": 0.3794335722923279,
|
| 10507 |
+
"learning_rate": 2.4338624338624338e-05,
|
| 10508 |
+
"loss": 0.6481,
|
| 10509 |
+
"step": 1500
|
| 10510 |
}
|
| 10511 |
],
|
| 10512 |
"logging_steps": 1,
|
| 10513 |
+
"max_steps": 1706,
|
| 10514 |
"num_input_tokens_seen": 0,
|
| 10515 |
"num_train_epochs": 2,
|
| 10516 |
"save_steps": 100,
|
| 10517 |
+
"total_flos": 2.2263110020544594e+18,
|
| 10518 |
"train_batch_size": 4,
|
| 10519 |
"trial_name": null,
|
| 10520 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5048
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33d3224c7da2f806a2a4179942e95de4bae50529784366269bad80e18dc944bd
|
| 3 |
size 5048
|