Hugging Face's logo

zenlm
/
zen-eco-4b-instruct

Text Generation
Transformers
Safetensors
GGUF
English
qwen3
zen
zenlm
hanzo
instruct
efficient
conversational
text-generation-inference
Model card Files
xet
 Community
zeekay commited on
Commit
8866a36
·
verified ·
1 Parent(s): c98ebac

Upload mlx/config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. mlx/config.json +54 -13
mlx/config.json CHANGED
@@ -1,20 +1,61 @@
1
  {
2
  "architectures": [
3
- "Qwen2ForCausalLM"
4
  ],
 
5
  "attention_dropout": 0.0,
6
- "bos_token_id": 151643,
7
  "eos_token_id": 151645,
 
8
  "hidden_act": "silu",
9
- "hidden_size": 2048,
10
  "initializer_range": 0.02,
11
- "intermediate_size": 11008,
12
- "max_position_embeddings": 32768,
13
- "max_window_layers": 70,
14
- "model_type": "qwen2",
15
- "num_attention_heads": 16,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
  "num_hidden_layers": 36,
17
- "num_key_value_heads": 2,
 
18
  "quantization": {
19
  "group_size": 64,
20
  "bits": 4,
@@ -26,11 +67,11 @@
26
  "mode": "affine"
27
  },
28
  "rms_norm_eps": 1e-06,
29
- "rope_theta": 1000000.0,
30
- "sliding_window": 32768,
 
31
  "tie_word_embeddings": true,
32
- "torch_dtype": "bfloat16",
33
- "transformers_version": "4.43.1",
34
  "use_cache": true,
35
  "use_sliding_window": false,
36
  "vocab_size": 151936
 
1
  {
2
  "architectures": [
3
+ "Qwen3ForCausalLM"
4
  ],
5
+ "attention_bias": false,
6
  "attention_dropout": 0.0,
7
+ "dtype": "float16",
8
  "eos_token_id": 151645,
9
+ "head_dim": 128,
10
  "hidden_act": "silu",
11
+ "hidden_size": 2560,
12
  "initializer_range": 0.02,
13
+ "intermediate_size": 9728,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention"
51
+ ],
52
+ "max_position_embeddings": 262144,
53
+ "max_window_layers": 36,
54
+ "model_type": "qwen3",
55
+ "num_attention_heads": 32,
56
  "num_hidden_layers": 36,
57
+ "num_key_value_heads": 8,
58
+ "pad_token_id": 151643,
59
  "quantization": {
60
  "group_size": 64,
61
  "bits": 4,
 
67
  "mode": "affine"
68
  },
69
  "rms_norm_eps": 1e-06,
70
+ "rope_scaling": null,
71
+ "rope_theta": 5000000,
72
+ "sliding_window": null,
73
  "tie_word_embeddings": true,
74
+ "transformers_version": "4.56.1",
 
75
  "use_cache": true,
76
  "use_sliding_window": false,
77
  "vocab_size": 151936