apepkuss79
commited on
Commit
•
3284a54
1
Parent(s):
95c43d6
Upload config.json with huggingface_hub
Browse files- config.json +31 -0
config.json
ADDED
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "starcoder2-instruct-15b-v0.1",
|
3 |
+
"architectures": [
|
4 |
+
"Starcoder2ForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 0,
|
8 |
+
"embedding_dropout": 0.0,
|
9 |
+
"eos_token_id": 0,
|
10 |
+
"hidden_act": "gelu_pytorch_tanh",
|
11 |
+
"hidden_size": 6144,
|
12 |
+
"initializer_range": 0.01275,
|
13 |
+
"intermediate_size": 24576,
|
14 |
+
"max_position_embeddings": 16384,
|
15 |
+
"mlp_type": "default",
|
16 |
+
"model_type": "starcoder2",
|
17 |
+
"norm_epsilon": 1e-05,
|
18 |
+
"norm_type": "layer_norm",
|
19 |
+
"num_attention_heads": 48,
|
20 |
+
"num_hidden_layers": 40,
|
21 |
+
"num_key_value_heads": 4,
|
22 |
+
"residual_dropout": 0.0,
|
23 |
+
"rope_theta": 100000,
|
24 |
+
"sliding_window": 4096,
|
25 |
+
"tie_word_embeddings": false,
|
26 |
+
"torch_dtype": "bfloat16",
|
27 |
+
"transformers_version": "4.39.0.dev0",
|
28 |
+
"use_bias": true,
|
29 |
+
"use_cache": true,
|
30 |
+
"vocab_size": 49152
|
31 |
+
}
|