Upload folder using huggingface_hub
Browse files- config.json +38 -0
- conversion_info.json +17 -0
- model.safetensors +3 -0
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen3MoeForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"model_type": "qwen3_moe",
|
| 6 |
+
"vocab_size": 151936,
|
| 7 |
+
"hidden_size": 1536,
|
| 8 |
+
"intermediate_size": 8960,
|
| 9 |
+
"moe_intermediate_size": 768,
|
| 10 |
+
"num_hidden_layers": 28,
|
| 11 |
+
"num_attention_heads": 12,
|
| 12 |
+
"num_key_value_heads": 2,
|
| 13 |
+
"head_dim": 64,
|
| 14 |
+
"hidden_act": "silu",
|
| 15 |
+
"max_position_embeddings": 32768,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"rms_norm_eps": 1e-06,
|
| 18 |
+
"use_cache": true,
|
| 19 |
+
"tie_word_embeddings": true,
|
| 20 |
+
"rope_theta": 1000000.0,
|
| 21 |
+
"rope_scaling": null,
|
| 22 |
+
"attention_bias": false,
|
| 23 |
+
"attention_dropout": 0.0,
|
| 24 |
+
"use_sliding_window": false,
|
| 25 |
+
"sliding_window": null,
|
| 26 |
+
"max_window_layers": 28,
|
| 27 |
+
"decoder_sparse_step": 1,
|
| 28 |
+
"num_experts": 288,
|
| 29 |
+
"num_experts_per_tok": 6,
|
| 30 |
+
"norm_topk_prob": true,
|
| 31 |
+
"output_router_logits": false,
|
| 32 |
+
"router_aux_loss_coef": 0.001,
|
| 33 |
+
"mlp_only_layers": [],
|
| 34 |
+
"torch_dtype": "bfloat16",
|
| 35 |
+
"transformers_version": "4.51.0",
|
| 36 |
+
"bos_token_id": 151643,
|
| 37 |
+
"eos_token_id": 151645
|
| 38 |
+
}
|
conversion_info.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"method": "drop_upcycling",
|
| 3 |
+
"drop_ratio": 0.66,
|
| 4 |
+
"num_experts_per_layer": 288,
|
| 5 |
+
"base_experts_from_ffn": 12,
|
| 6 |
+
"replication_factor": 24,
|
| 7 |
+
"total_experts": 8064,
|
| 8 |
+
"source_ffn_dim": 8960,
|
| 9 |
+
"target_expert_dim": 768,
|
| 10 |
+
"layers": 28,
|
| 11 |
+
"total_params": 28938036224,
|
| 12 |
+
"total_params_calculated": 28860965888,
|
| 13 |
+
"active_params_per_forward": 917464064,
|
| 14 |
+
"activation_ratio": 0.03170443415365876,
|
| 15 |
+
"expert_params": 3538944,
|
| 16 |
+
"total_expert_params": 28538044416
|
| 17 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f7cffecaeec34ccb6b1d2c98cbba814590ace14802977f38d937500a11ea844
|
| 3 |
+
size 57879156016
|