rkazants commited on
Commit
8028d11
·
verified ·
1 Parent(s): b9b97e5

Upload 3 files

Browse files
Files changed (3) hide show
  1. config.json +5 -5
  2. generation_config.json +1 -1
  3. model.safetensors +2 -2
config.json CHANGED
@@ -15,11 +15,11 @@
15
  },
16
  "dtype": "float32",
17
  "eos_token_id": null,
18
- "head_dim": 4,
19
  "hidden_act": "gelu_pytorch_tanh",
20
- "hidden_size": 8,
21
  "initializer_range": 0.02,
22
- "intermediate_size": 32,
23
  "layer_types": [
24
  "full_attention",
25
  "full_attention"
@@ -27,7 +27,7 @@
27
  "max_position_embeddings": 131072,
28
  "max_window_layers": 28,
29
  "model_type": "qwen3",
30
- "num_attention_heads": 2,
31
  "num_hidden_layers": 2,
32
  "num_key_value_heads": 1,
33
  "pad_token_id": null,
@@ -38,7 +38,7 @@
38
  },
39
  "sliding_window": null,
40
  "tie_word_embeddings": false,
41
- "transformers_version": "5.2.0",
42
  "use_cache": true,
43
  "use_sliding_window": false,
44
  "vocab_size": 262144
 
15
  },
16
  "dtype": "float32",
17
  "eos_token_id": null,
18
+ "head_dim": 8,
19
  "hidden_act": "gelu_pytorch_tanh",
20
+ "hidden_size": 32,
21
  "initializer_range": 0.02,
22
+ "intermediate_size": 64,
23
  "layer_types": [
24
  "full_attention",
25
  "full_attention"
 
27
  "max_position_embeddings": 131072,
28
  "max_window_layers": 28,
29
  "model_type": "qwen3",
30
+ "num_attention_heads": 4,
31
  "num_hidden_layers": 2,
32
  "num_key_value_heads": 1,
33
  "pad_token_id": null,
 
38
  },
39
  "sliding_window": null,
40
  "tie_word_embeddings": false,
41
+ "transformers_version": "5.5.0",
42
  "use_cache": true,
43
  "use_sliding_window": false,
44
  "vocab_size": 262144
generation_config.json CHANGED
@@ -2,6 +2,6 @@
2
  "_from_model_config": true,
3
  "output_attentions": false,
4
  "output_hidden_states": false,
5
- "transformers_version": "5.2.0",
6
  "use_cache": true
7
  }
 
2
  "_from_model_config": true,
3
  "output_attentions": false,
4
  "output_hidden_states": false,
5
+ "transformers_version": "5.5.0",
6
  "use_cache": true
7
  }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b423975c5051beb30f8328dfceab5645470cd8debb99cef358dfaf49a8182ca5
3
- size 10704
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70cf6bbac674a7f172a87e74f8880b9523e2570a3e946465ecaf08fb3eeefd72
3
+ size 81048