Model save
Browse files- README.md +11 -5
- adapter_config.json +2 -2
- adapter_model.safetensors +1 -1
- all_results.json +19 -6
- eval_results.json +16 -0
- train_results.json +6 -6
- trainer_state.json +37 -9
- training_args.bin +1 -1
README.md
CHANGED
@@ -2,13 +2,9 @@
|
|
2 |
license: mit
|
3 |
library_name: peft
|
4 |
tags:
|
5 |
-
- alignment-handbook
|
6 |
-
- generated_from_trainer
|
7 |
- trl
|
8 |
- dpo
|
9 |
- generated_from_trainer
|
10 |
-
datasets:
|
11 |
-
- HuggingFaceH4/ultrafeedback_binarized
|
12 |
base_model: microsoft/phi-2
|
13 |
model-index:
|
14 |
- name: phi-2-gpo-iter-0
|
@@ -20,7 +16,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
20 |
|
21 |
# phi-2-gpo-iter-0
|
22 |
|
23 |
-
This model is a fine-tuned version of [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
|
25 |
## Model description
|
26 |
|
|
|
2 |
license: mit
|
3 |
library_name: peft
|
4 |
tags:
|
|
|
|
|
5 |
- trl
|
6 |
- dpo
|
7 |
- generated_from_trainer
|
|
|
|
|
8 |
base_model: microsoft/phi-2
|
9 |
model-index:
|
10 |
- name: phi-2-gpo-iter-0
|
|
|
16 |
|
17 |
# phi-2-gpo-iter-0
|
18 |
|
19 |
+
This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on the None dataset.
|
20 |
+
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 0.0103
|
22 |
+
- Rewards/chosen: 0.0006
|
23 |
+
- Rewards/rejected: 0.0005
|
24 |
+
- Rewards/accuracies: 0.4945
|
25 |
+
- Rewards/margins: 0.0001
|
26 |
+
- Logps/rejected: -233.4960
|
27 |
+
- Logps/chosen: -256.5553
|
28 |
+
- Logits/rejected: 0.8946
|
29 |
+
- Logits/chosen: 0.8373
|
30 |
|
31 |
## Model description
|
32 |
|
adapter_config.json
CHANGED
@@ -20,9 +20,9 @@
|
|
20 |
"revision": null,
|
21 |
"target_modules": [
|
22 |
"dense",
|
23 |
-
"
|
24 |
"q_proj",
|
25 |
-
"
|
26 |
],
|
27 |
"task_type": "CAUSAL_LM"
|
28 |
}
|
|
|
20 |
"revision": null,
|
21 |
"target_modules": [
|
22 |
"dense",
|
23 |
+
"k_proj",
|
24 |
"q_proj",
|
25 |
+
"v_proj"
|
26 |
],
|
27 |
"task_type": "CAUSAL_LM"
|
28 |
}
|
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41977616
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:279aa434e1b02b80bc6174c710c9fcad6a53203ae32da4a203d6a9e813474420
|
3 |
size 41977616
|
all_results.json
CHANGED
@@ -1,8 +1,21 @@
|
|
1 |
{
|
2 |
-
"epoch": 1.
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 1.99,
|
3 |
+
"eval_logits/chosen": 0.8372604250907898,
|
4 |
+
"eval_logits/rejected": 0.8945639133453369,
|
5 |
+
"eval_logps/chosen": -256.5552978515625,
|
6 |
+
"eval_logps/rejected": -233.49595642089844,
|
7 |
+
"eval_loss": 0.01026910264045,
|
8 |
+
"eval_rewards/accuracies": 0.4945000112056732,
|
9 |
+
"eval_rewards/chosen": 0.0006274624029174447,
|
10 |
+
"eval_rewards/margins": 0.00010108990682056174,
|
11 |
+
"eval_rewards/rejected": 0.0005263725761324167,
|
12 |
+
"eval_runtime": 709.1992,
|
13 |
+
"eval_samples": 2000,
|
14 |
+
"eval_samples_per_second": 2.82,
|
15 |
+
"eval_steps_per_second": 0.705,
|
16 |
+
"train_loss": 0.001956978782771095,
|
17 |
+
"train_runtime": 290.9594,
|
18 |
+
"train_samples": 61135,
|
19 |
+
"train_samples_per_second": 6.874,
|
20 |
+
"train_steps_per_second": 0.426
|
21 |
}
|
eval_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.99,
|
3 |
+
"eval_logits/chosen": 0.8372604250907898,
|
4 |
+
"eval_logits/rejected": 0.8945639133453369,
|
5 |
+
"eval_logps/chosen": -256.5552978515625,
|
6 |
+
"eval_logps/rejected": -233.49595642089844,
|
7 |
+
"eval_loss": 0.01026910264045,
|
8 |
+
"eval_rewards/accuracies": 0.4945000112056732,
|
9 |
+
"eval_rewards/chosen": 0.0006274624029174447,
|
10 |
+
"eval_rewards/margins": 0.00010108990682056174,
|
11 |
+
"eval_rewards/rejected": 0.0005263725761324167,
|
12 |
+
"eval_runtime": 709.1992,
|
13 |
+
"eval_samples": 2000,
|
14 |
+
"eval_samples_per_second": 2.82,
|
15 |
+
"eval_steps_per_second": 0.705
|
16 |
+
}
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"epoch": 1.
|
3 |
-
"train_loss": 0.
|
4 |
-
"train_runtime":
|
5 |
-
"train_samples":
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second": 0.
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 1.99,
|
3 |
+
"train_loss": 0.001956978782771095,
|
4 |
+
"train_runtime": 290.9594,
|
5 |
+
"train_samples": 61135,
|
6 |
+
"train_samples_per_second": 6.874,
|
7 |
+
"train_steps_per_second": 0.426
|
8 |
}
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 1.
|
5 |
"eval_steps": 100,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -163,17 +163,45 @@
|
|
163 |
"step": 100
|
164 |
},
|
165 |
{
|
166 |
-
"epoch": 1.
|
167 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
168 |
"total_flos": 0.0,
|
169 |
-
"train_loss": 0.
|
170 |
-
"train_runtime":
|
171 |
-
"train_samples_per_second":
|
172 |
-
"train_steps_per_second": 0.
|
173 |
}
|
174 |
],
|
175 |
"logging_steps": 10,
|
176 |
-
"max_steps":
|
177 |
"num_input_tokens_seen": 0,
|
178 |
"num_train_epochs": 2,
|
179 |
"save_steps": 100,
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.992,
|
5 |
"eval_steps": 100,
|
6 |
+
"global_step": 124,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
163 |
"step": 100
|
164 |
},
|
165 |
{
|
166 |
+
"epoch": 1.77,
|
167 |
+
"learning_rate": 1.937002879188285e-07,
|
168 |
+
"logits/chosen": 0.8532508611679077,
|
169 |
+
"logits/rejected": 0.8138865232467651,
|
170 |
+
"logps/chosen": -175.2100830078125,
|
171 |
+
"logps/rejected": -158.42974853515625,
|
172 |
+
"loss": 0.0102,
|
173 |
+
"rewards/accuracies": 0.48124998807907104,
|
174 |
+
"rewards/chosen": -0.0008482037228532135,
|
175 |
+
"rewards/margins": 0.00030287052504718304,
|
176 |
+
"rewards/rejected": -0.0011510740732774138,
|
177 |
+
"step": 110
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 1.93,
|
181 |
+
"learning_rate": 1.6003680950742728e-08,
|
182 |
+
"logits/chosen": 0.8294161558151245,
|
183 |
+
"logits/rejected": 0.826370120048523,
|
184 |
+
"logps/chosen": -158.0736083984375,
|
185 |
+
"logps/rejected": -146.1743927001953,
|
186 |
+
"loss": 0.01,
|
187 |
+
"rewards/accuracies": 0.48750001192092896,
|
188 |
+
"rewards/chosen": -0.0006492567481473088,
|
189 |
+
"rewards/margins": 0.0010562509996816516,
|
190 |
+
"rewards/rejected": -0.0017055077478289604,
|
191 |
+
"step": 120
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 1.99,
|
195 |
+
"step": 124,
|
196 |
"total_flos": 0.0,
|
197 |
+
"train_loss": 0.001956978782771095,
|
198 |
+
"train_runtime": 290.9594,
|
199 |
+
"train_samples_per_second": 6.874,
|
200 |
+
"train_steps_per_second": 0.426
|
201 |
}
|
202 |
],
|
203 |
"logging_steps": 10,
|
204 |
+
"max_steps": 124,
|
205 |
"num_input_tokens_seen": 0,
|
206 |
"num_train_epochs": 2,
|
207 |
"save_steps": 100,
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 5816
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8eae090f69235327d3649619a5e45322d1fd2adadd2e607fcfed49b1f3988ff4
|
3 |
size 5816
|