C10X commited on
Commit
c2f5196
·
verified ·
1 Parent(s): 9c8247c

Upload 8 files

Browse files
config.json ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3MoeForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 50256,
8
+ "decoder_sparse_step": 1,
9
+ "dtype": "bfloat16",
10
+ "eos_token_id": 50256,
11
+ "head_dim": 32,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 128,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 1024,
16
+ "layer_types": [
17
+ "sliding_attention",
18
+ "sliding_attention",
19
+ "full_attention",
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "full_attention",
23
+ "sliding_attention",
24
+ "sliding_attention"
25
+ ],
26
+ "max_position_embeddings": 8192,
27
+ "max_window_layers": 6,
28
+ "mlp_only_layers": [],
29
+ "model_type": "qwen3_moe",
30
+ "moe_intermediate_size": 1024,
31
+ "norm_topk_prob": true,
32
+ "num_attention_heads": 4,
33
+ "num_experts": 64,
34
+ "num_experts_per_tok": 2,
35
+ "num_hidden_layers": 8,
36
+ "num_key_value_heads": 2,
37
+ "output_router_logits": false,
38
+ "pad_token_id": 50256,
39
+ "query_pre_attn_scalar": 256,
40
+ "rms_norm_eps": 1e-06,
41
+ "rope_scaling": null,
42
+ "rope_theta": 500000,
43
+ "router_aux_loss_coef": 0.001,
44
+ "sliding_window": 512,
45
+ "tie_word_embeddings": true,
46
+ "transformers_version": "4.57.6",
47
+ "use_cache": false,
48
+ "use_sliding_window": true,
49
+ "vocab_size": 50257
50
+ }
generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 50256,
3
+ "eos_token_id": 50256,
4
+ "pad_token_id": 50256,
5
+ "max_new_tokens": 256,
6
+ "do_sample": true,
7
+ "temperature": 0.7,
8
+ "top_p": 0.9
9
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
metadata.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "Qwen3-5M-MoE-2exp-active",
3
+ "model_type": "Qwen3MoeForCausalLM",
4
+ "tokenizer": "gpt2",
5
+ "dtype": "bfloat16",
6
+ "vocab_size": 50257,
7
+ "hidden_size": 128,
8
+ "num_layers": 8,
9
+ "num_attention_heads": 4,
10
+ "num_key_value_heads": 2,
11
+ "head_dim": 32,
12
+ "moe_intermediate_size": 1024,
13
+ "num_experts": 64,
14
+ "num_experts_per_tok": 2,
15
+ "sliding_window": 512,
16
+ "max_position_embeddings": 8192,
17
+ "rope_theta": 500000,
18
+ "layer_types": [
19
+ "sliding_attention",
20
+ "sliding_attention",
21
+ "full_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "full_attention",
25
+ "sliding_attention",
26
+ "sliding_attention"
27
+ ],
28
+ "max_window_layers": 6,
29
+ "parameters_total": 208220928,
30
+ "parameters_active": 13186816,
31
+ "active_ratio": 0.03125,
32
+ "positional_encoding": "rope",
33
+ "normalization": "rmsnorm",
34
+ "activation": "swiglu",
35
+ "tie_word_embeddings": true
36
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<|endoftext|>",
3
+ "eos_token": "<|endoftext|>",
4
+ "pad_token": "<|endoftext|>",
5
+ "unk_token": "<|endoftext|>"
6
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "50256": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ }
12
+ },
13
+ "bos_token": "<|endoftext|>",
14
+ "clean_up_tokenization_spaces": false,
15
+ "eos_token": "<|endoftext|>",
16
+ "extra_special_tokens": {},
17
+ "model_max_length": 1024,
18
+ "pad_token": "<|endoftext|>",
19
+ "tokenizer_class": "GPT2Tokenizer",
20
+ "unk_token": "<|endoftext|>"
21
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff