whizzzzkid
commited on
Commit
•
3dccca6
1
Parent(s):
630dc22
Upload StableLmForCausalLM
Browse files- config.json +3 -2
- generation_config.json +2 -1
- model.safetensors +1 -1
config.json
CHANGED
@@ -5,8 +5,7 @@
|
|
5 |
],
|
6 |
"attention_dropout": 0.0,
|
7 |
"bos_token_id": 100289,
|
8 |
-
"eos_token_id":
|
9 |
-
"eos_token_id_was": 100290,
|
10 |
"hidden_act": "silu",
|
11 |
"hidden_dropout": 0.0,
|
12 |
"hidden_size": 2048,
|
@@ -19,12 +18,14 @@
|
|
19 |
"num_hidden_layers": 24,
|
20 |
"num_key_value_heads": 32,
|
21 |
"partial_rotary_factor": 0.25,
|
|
|
22 |
"rope_scaling": null,
|
23 |
"rope_theta": 10000,
|
24 |
"tie_word_embeddings": false,
|
25 |
"torch_dtype": "bfloat16",
|
26 |
"transformers_version": "4.38.2",
|
27 |
"use_cache": true,
|
|
|
28 |
"use_qkv_bias": true,
|
29 |
"vocab_size": 100352
|
30 |
}
|
|
|
5 |
],
|
6 |
"attention_dropout": 0.0,
|
7 |
"bos_token_id": 100289,
|
8 |
+
"eos_token_id": 100290,
|
|
|
9 |
"hidden_act": "silu",
|
10 |
"hidden_dropout": 0.0,
|
11 |
"hidden_size": 2048,
|
|
|
18 |
"num_hidden_layers": 24,
|
19 |
"num_key_value_heads": 32,
|
20 |
"partial_rotary_factor": 0.25,
|
21 |
+
"qk_layernorm": false,
|
22 |
"rope_scaling": null,
|
23 |
"rope_theta": 10000,
|
24 |
"tie_word_embeddings": false,
|
25 |
"torch_dtype": "bfloat16",
|
26 |
"transformers_version": "4.38.2",
|
27 |
"use_cache": true,
|
28 |
+
"use_parallel_residual": false,
|
29 |
"use_qkv_bias": true,
|
30 |
"vocab_size": 100352
|
31 |
}
|
generation_config.json
CHANGED
@@ -2,5 +2,6 @@
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 100289,
|
4 |
"eos_token_id": 100290,
|
5 |
-
"transformers_version": "4.38.2"
|
|
|
6 |
}
|
|
|
2 |
"_from_model_config": true,
|
3 |
"bos_token_id": 100289,
|
4 |
"eos_token_id": 100290,
|
5 |
+
"transformers_version": "4.38.2",
|
6 |
+
"use_cache": false
|
7 |
}
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3289069520
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2d67caa172654ea3fb257cec537de3c77b34acb1685fabf01a41213883305087
|
3 |
size 3289069520
|