Hugging Face's logo

Synthyra
/
Profluent-E1-600M

Fill-Mask
Transformers
Safetensors
E1
custom_code
Model card Files
xet
 Community
lhallee commited on
Commit
f174801
·
verified ·
1 Parent(s): 659e32b

Upload config

Browse files
Files changed (1) hide show
  1. config.json +40 -40
config.json CHANGED
@@ -1,40 +1,40 @@
1
- {
2
- "architectures": [
3
- "E1ForMaskedLM"
4
- ],
5
- "attn_backend": "sdpa",
6
- "auto_map": {
7
- "AutoConfig": "modeling_e1.E1Config",
8
- "AutoModel": "modeling_e1.E1Model",
9
- "AutoModelForMaskedLM": "modeling_e1.E1ForMaskedLM",
10
- "AutoModelForSequenceClassification": "modeling_e1.E1ForSequenceClassification",
11
- "AutoModelForTokenClassification": "modeling_e1.E1ForTokenClassification"
12
- },
13
- "bos_token_id": 1,
14
- "clip_qkv": 8,
15
- "dtype": "bfloat16",
16
- "eos_token_id": 2,
17
- "gated_mlp": true,
18
- "global_attention_every_n_layers": 3,
19
- "gradient_checkpointing": false,
20
- "hidden_act": "silu",
21
- "hidden_size": 1280,
22
- "initializer_range": 0.02,
23
- "intermediate_size": 3840,
24
- "max_num_positions_global": 65536,
25
- "max_num_positions_within_seq": 8192,
26
- "max_num_sequences": 512,
27
- "model_type": "E1",
28
- "no_ffn_gradient_checkpointing": true,
29
- "num_attention_heads": 20,
30
- "num_hidden_layers": 30,
31
- "num_key_value_heads": 20,
32
- "pad_token_id": 0,
33
- "rms_norm_eps": 1e-05,
34
- "rope_theta_global": 500000.0,
35
- "rope_theta_within_seq": 10000.0,
36
- "tie_word_embeddings": false,
37
- "transformers_version": "4.48.1",
38
- "use_cache": true,
39
- "vocab_size": 34
40
- }
 
1
+ {
2
+ "architectures": [
3
+ "E1ForMaskedLM"
4
+ ],
5
+ "attn_backend": "sdpa",
6
+ "auto_map": {
7
+ "AutoConfig": "modeling_e1.E1Config",
8
+ "AutoModel": "modeling_e1.E1Model",
9
+ "AutoModelForMaskedLM": "modeling_e1.E1ForMaskedLM",
10
+ "AutoModelForSequenceClassification": "modeling_e1.E1ForSequenceClassification",
11
+ "AutoModelForTokenClassification": "modeling_e1.E1ForTokenClassification"
12
+ },
13
+ "bos_token_id": 1,
14
+ "clip_qkv": 8,
15
+ "dtype": "bfloat16",
16
+ "eos_token_id": 2,
17
+ "gated_mlp": true,
18
+ "global_attention_every_n_layers": 3,
19
+ "gradient_checkpointing": false,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 1280,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 3840,
24
+ "max_num_positions_global": 65536,
25
+ "max_num_positions_within_seq": 8192,
26
+ "max_num_sequences": 512,
27
+ "model_type": "E1",
28
+ "no_ffn_gradient_checkpointing": true,
29
+ "num_attention_heads": 20,
30
+ "num_hidden_layers": 30,
31
+ "num_key_value_heads": 20,
32
+ "pad_token_id": 0,
33
+ "rms_norm_eps": 1e-05,
34
+ "rope_theta_global": 500000.0,
35
+ "rope_theta_within_seq": 10000.0,
36
+ "tie_word_embeddings": false,
37
+ "transformers_version": "4.57.6",
38
+ "use_cache": true,
39
+ "vocab_size": 34
40
+ }