JiayouZhangGenbio commited on
Commit
04203c6
·
verified ·
1 Parent(s): eaae91b

Upload folder using huggingface_hub

Browse files
config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_linear_bias": true,
3
+ "architectures": [
4
+ "FM4BioForMaskedLM"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "experts_per_token": 2,
8
+ "hidden_act": "swiglu",
9
+ "hidden_dropout_prob": 0.0,
10
+ "hidden_size": 2304,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 7680,
13
+ "layer_norm_eps": 1e-05,
14
+ "max_position_embeddings": 1024,
15
+ "model_type": "fm4bio",
16
+ "moe": true,
17
+ "normalization_type": "RMSNorm",
18
+ "num_attention_heads": 36,
19
+ "num_experts": 8,
20
+ "num_hidden_layers": 36,
21
+ "output_vocab_size": 512,
22
+ "pad_token_id": 0,
23
+ "position_embedding_type": "rope",
24
+ "rotary_percent": 1.0,
25
+ "seq_len_interpolation_factor": null,
26
+ "tie_word_embeddings": false,
27
+ "torch_dtype": "float32",
28
+ "transformers_version": "4.38.0.dev0",
29
+ "type_vocab_size": 2,
30
+ "use_cache": true,
31
+ "use_lm_head": false,
32
+ "vocab_size": 128
33
+ }
generation_config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "pad_token_id": 0,
4
+ "transformers_version": "4.38.0.dev0"
5
+ }
pytorch_model-00001-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fb93bea5f60b0ce9599c26989e9be40ad285ffde64bddb5bedbfbd3cd1669bf7
3
+ size 4929373628
pytorch_model-00002-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0f9d566d4c751b95db4a9bdf3826dc047e713d89fac31291ad4d8726a19ae1f
3
+ size 4928193593
pytorch_model-00003-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f16a854ec54e692756251199daaf41e9bc1819b9afc25bcedb2902623fc598d
3
+ size 4928193521
pytorch_model-00004-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26f4ec822eadc0f8efad87209ec17d115bbbc0857bdf44b3d951ea7edaf4eb82
3
+ size 4928193585
pytorch_model-00005-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2c210159bf433d78e341b3fd00d296d591cd8efe7aaa15a7f8e5a46e2abd629
3
+ size 4984746078
pytorch_model-00006-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9f730aa72a99afcb76e7bdf6502e340064c8732daed5db9dfe77d58e213afc3
3
+ size 4998982519
pytorch_model-00007-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f17f624f2a858933df55f1677f0372a5f4d0635185806c98a0e3f4704e897bc9
3
+ size 4928193673
pytorch_model-00008-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e99e5c9f8d1e4faa4448ec92326f92a3a3d061bfc271f1eaeb75d99b66f47ca
3
+ size 4928193649
pytorch_model-00009-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b97b71231f8f1e4195530e4ff628ad59f2f25980d09852c4407e1d40c34b0463
3
+ size 4984746078
pytorch_model-00010-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da6f2549dee72a3fef807ea6b04348429c4bf3906e6eca9848052b1fe33deaf3
3
+ size 4998982555
pytorch_model-00011-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7de2f95018f3ec53baca50d65c9b860273db186550463fa50ce49c2dc85ea6a4
3
+ size 4928193721
pytorch_model-00012-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69531b500c8f3ebf98910e903fde86d31c9e9fa4068970bc9bd65ab3f7b388c5
3
+ size 4928193649
pytorch_model-00013-of-00013.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a29cc85868bcd5d6e0f26fd40e10c07700d5ca5a5f1e7338a954dd004327b50
3
+ size 4847854626
pytorch_model.bin.index.json ADDED
The diff for this file is too large to render. See raw diff