SeerAttention commited on
Commit
bb2ad01
·
verified ·
1 Parent(s): 922faa6

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. attn_gate_weights.pth +3 -0
  2. config.json +38 -0
attn_gate_weights.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:413f1a792ed0d5cd183e39ff40b80cb5b3c25303f62ba0fff5e49259e732d4c0
3
+ size 66101646
config.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_attn_implementation_autoset": true,
3
+ "architectures": [
4
+ "Qwen3ForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "base_model": "Qwen/Qwen3-8B",
9
+ "bos_token_id": 151643,
10
+ "eos_token_id": 151645,
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 4096,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 12288,
16
+ "max_position_embeddings": 40960,
17
+ "max_window_layers": 36,
18
+ "model_type": "qwen3",
19
+ "num_attention_heads": 32,
20
+ "num_hidden_layers": 36,
21
+ "num_key_value_heads": 8,
22
+ "rms_norm_eps": 1e-06,
23
+ "rope_scaling": null,
24
+ "rope_theta": 1000000,
25
+ "seerattn_gate_block_size": 64,
26
+ "seerattn_gate_hidden_size": 128,
27
+ "seerattn_k_seq_pooling_type": "Kmaxminavg",
28
+ "seerattn_q_head_pooling_type": "Qproj",
29
+ "seerattn_use_qk_norm": true,
30
+ "seerattn_use_rope": true,
31
+ "sliding_window": null,
32
+ "tie_word_embeddings": false,
33
+ "torch_dtype": "bfloat16",
34
+ "transformers_version": "4.51.0",
35
+ "use_cache": true,
36
+ "use_sliding_window": false,
37
+ "vocab_size": 151936
38
+ }