PatrickHaller
commited on
Commit
•
ed6f284
1
Parent(s):
bf938ec
Upload config.json with huggingface_hub
Browse files- config.json +35 -0
config.json
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"HGRN2ForCausalLM"
|
4 |
+
],
|
5 |
+
"attn_mode": "chunk",
|
6 |
+
"bos_token_id": 1,
|
7 |
+
"conv_size": 4,
|
8 |
+
"elementwise_affine": true,
|
9 |
+
"eos_token_id": 2,
|
10 |
+
"expand_ratio": 128,
|
11 |
+
"fuse_cross_entropy": true,
|
12 |
+
"hidden_act": "swish",
|
13 |
+
"hidden_ratio": 4,
|
14 |
+
"hidden_size": 2048,
|
15 |
+
"initializer_range": 0.02,
|
16 |
+
"intermediate_size": null,
|
17 |
+
"max_position_embeddings": 2048,
|
18 |
+
"model_type": "hgrn2",
|
19 |
+
"norm_eps": 1e-06,
|
20 |
+
"num_heads": null,
|
21 |
+
"num_hidden_layers": 18,
|
22 |
+
"tie_word_embeddings": false,
|
23 |
+
"torch_dtype": "float32",
|
24 |
+
"transformers_version": "4.44.1",
|
25 |
+
"use_cache": true,
|
26 |
+
"use_lower_bound": true,
|
27 |
+
"use_short_conv": false,
|
28 |
+
"vocab_size": 16000,
|
29 |
+
"auto_map": {
|
30 |
+
"AutoConfig": "configuration_hgrn2.HGRN2Config",
|
31 |
+
"AutoModelForCausalLM": "modeling_hgrn2.HGRN2ForCausalLM",
|
32 |
+
"AutoModel": "modeling_hgrn2.HGRN2Model",
|
33 |
+
"AutoModelForSequenceClassification": "modeling_hgrn2.HGRN2ForSequenceClassification"
|
34 |
+
}
|
35 |
+
}
|