edit config
Browse files- config.json +13 -12
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"architectures": [
|
4 |
"MistralForCausalLM"
|
5 |
],
|
@@ -19,14 +19,8 @@
|
|
19 |
"num_attention_heads": 32,
|
20 |
"num_hidden_layers": 32,
|
21 |
"num_key_value_heads": 8,
|
22 |
-
"
|
23 |
-
|
24 |
-
"group_size": 128,
|
25 |
-
"modules_to_not_convert": null,
|
26 |
-
"quant_method": "awq",
|
27 |
-
"version": "gemm",
|
28 |
-
"zero_point": true
|
29 |
-
},
|
30 |
"rms_norm_eps": 1e-05,
|
31 |
"rope_scaling": {
|
32 |
"factor": 16.0,
|
@@ -38,7 +32,14 @@
|
|
38 |
"sliding_window": 131072,
|
39 |
"tie_word_embeddings": false,
|
40 |
"torch_dtype": "float16",
|
41 |
-
"transformers_version": "4.
|
42 |
"use_cache": true,
|
43 |
-
"vocab_size": 32000
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "/workspace/process/nousresearch_yarn-mistral-7b-128k/source",
|
3 |
"architectures": [
|
4 |
"MistralForCausalLM"
|
5 |
],
|
|
|
19 |
"num_attention_heads": 32,
|
20 |
"num_hidden_layers": 32,
|
21 |
"num_key_value_heads": 8,
|
22 |
+
"pad_token_id": 0,
|
23 |
+
"pretraining_tp": 1,
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
"rms_norm_eps": 1e-05,
|
25 |
"rope_scaling": {
|
26 |
"factor": 16.0,
|
|
|
32 |
"sliding_window": 131072,
|
33 |
"tie_word_embeddings": false,
|
34 |
"torch_dtype": "float16",
|
35 |
+
"transformers_version": "4.35.0",
|
36 |
"use_cache": true,
|
37 |
+
"vocab_size": 32000,
|
38 |
+
"quantization_config": {
|
39 |
+
"quant_method": "awq",
|
40 |
+
"zero_point": true,
|
41 |
+
"group_size": 128,
|
42 |
+
"bits": 4,
|
43 |
+
"version": "gemm"
|
44 |
+
}
|
45 |
+
}
|