HDCharles commited on
Commit
72764fb
·
verified ·
1 Parent(s): 0af8f5c

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. config.json +6 -6
  2. model.safetensors +2 -2
config.json CHANGED
@@ -7,25 +7,25 @@
7
  "decoder_sparse_step": 1,
8
  "dtype": "float32",
9
  "hidden_act": "silu",
10
- "hidden_size": 256,
11
  "initializer_range": 0.02,
12
- "intermediate_size": 256,
13
  "max_position_embeddings": 32768,
14
  "mlp_only_layers": [],
15
  "model_type": "qwen3_moe",
16
- "moe_intermediate_size": 256,
17
  "norm_topk_prob": false,
18
- "num_attention_heads": 16,
19
  "num_experts": 8,
20
  "num_experts_per_tok": 2,
21
  "num_hidden_layers": 6,
22
- "num_key_value_heads": 16,
23
  "output_router_logits": false,
24
  "rms_norm_eps": 1e-06,
25
  "rope_scaling": null,
26
  "rope_theta": 10000.0,
27
  "router_aux_loss_coef": 0.001,
28
- "shared_expert_intermediate_size": 256,
29
  "sliding_window": null,
30
  "tie_word_embeddings": true,
31
  "transformers_version": "4.57.1",
 
7
  "decoder_sparse_step": 1,
8
  "dtype": "float32",
9
  "hidden_act": "silu",
10
+ "hidden_size": 128,
11
  "initializer_range": 0.02,
12
+ "intermediate_size": 128,
13
  "max_position_embeddings": 32768,
14
  "mlp_only_layers": [],
15
  "model_type": "qwen3_moe",
16
+ "moe_intermediate_size": 128,
17
  "norm_topk_prob": false,
18
+ "num_attention_heads": 4,
19
  "num_experts": 8,
20
  "num_experts_per_tok": 2,
21
  "num_hidden_layers": 6,
22
+ "num_key_value_heads": 2,
23
  "output_router_logits": false,
24
  "rms_norm_eps": 1e-06,
25
  "rope_scaling": null,
26
  "rope_theta": 10000.0,
27
  "router_aux_loss_coef": 0.001,
28
+ "shared_expert_intermediate_size": 128,
29
  "sliding_window": null,
30
  "tie_word_embeddings": true,
31
  "transformers_version": "4.57.1",
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6f30f0db40aca3eebbc72c426c60be4d2c4f59c08e41fb6118f198d66c04b49d
3
- size 46245904
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c6200ba590e2490caff59b5cb31e18bd5ef11f81e1b66236349f8abf8a57ae9
3
+ size 11731960