comb-project commited on
Commit ·
b2e8d52
1
Parent(s): eb262a2
Upload weights
Browse files- config.json +80 -0
- generation_config.json +7 -0
- model-00001-of-00008.safetensors +3 -0
- model-00002-of-00008.safetensors +3 -0
- model-00003-of-00008.safetensors +3 -0
- model-00004-of-00008.safetensors +3 -0
- model-00005-of-00008.safetensors +3 -0
- model-00006-of-00008.safetensors +3 -0
- model-00007-of-00008.safetensors +3 -0
- model-00008-of-00008.safetensors +3 -0
- model.safetensors.index.json +0 -0
config.json
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"CombDeepSeekForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"chunk_token_index": 100002,
|
| 6 |
+
"cross_attention_layers": [
|
| 7 |
+
3,
|
| 8 |
+
7,
|
| 9 |
+
11,
|
| 10 |
+
15,
|
| 11 |
+
19,
|
| 12 |
+
23
|
| 13 |
+
],
|
| 14 |
+
"dtype": "bfloat16",
|
| 15 |
+
"model_type": "combdeepseek",
|
| 16 |
+
"num_hidden_layers": 33,
|
| 17 |
+
"pad_token_id": 100001,
|
| 18 |
+
"text_config": {
|
| 19 |
+
"architectures": [
|
| 20 |
+
"DeepseekV2ForCausalLM"
|
| 21 |
+
],
|
| 22 |
+
"attention_bias": false,
|
| 23 |
+
"attention_dropout": 0.0,
|
| 24 |
+
"auto_map": {
|
| 25 |
+
"AutoConfig": "configuration_deepseek.DeepseekV2Config",
|
| 26 |
+
"AutoModel": "modeling_deepseek.DeepseekV2Model",
|
| 27 |
+
"AutoModelForCausalLM": "modeling_deepseek.DeepseekV2ForCausalLM"
|
| 28 |
+
},
|
| 29 |
+
"aux_loss_alpha": 0.001,
|
| 30 |
+
"bos_token_id": 100000,
|
| 31 |
+
"dtype": "bfloat16",
|
| 32 |
+
"eos_token_id": 100001,
|
| 33 |
+
"first_k_dense_replace": 1,
|
| 34 |
+
"head_dim": 64,
|
| 35 |
+
"hidden_act": "silu",
|
| 36 |
+
"hidden_size": 2048,
|
| 37 |
+
"initializer_range": 0.02,
|
| 38 |
+
"intermediate_size": 10944,
|
| 39 |
+
"kv_lora_rank": 512,
|
| 40 |
+
"max_position_embeddings": 163840,
|
| 41 |
+
"mlp_bias": false,
|
| 42 |
+
"model_type": "deepseek_v2",
|
| 43 |
+
"moe_intermediate_size": 1408,
|
| 44 |
+
"moe_layer_freq": 1,
|
| 45 |
+
"n_group": 1,
|
| 46 |
+
"n_routed_experts": 64,
|
| 47 |
+
"n_shared_experts": 2,
|
| 48 |
+
"norm_topk_prob": false,
|
| 49 |
+
"num_attention_heads": 16,
|
| 50 |
+
"num_experts_per_tok": 6,
|
| 51 |
+
"num_hidden_layers": 33,
|
| 52 |
+
"num_key_value_heads": 16,
|
| 53 |
+
"pretraining_tp": 1,
|
| 54 |
+
"q_lora_rank": null,
|
| 55 |
+
"qk_nope_head_dim": 128,
|
| 56 |
+
"qk_rope_head_dim": 64,
|
| 57 |
+
"rms_norm_eps": 1e-06,
|
| 58 |
+
"rope_scaling": {
|
| 59 |
+
"beta_fast": 32.0,
|
| 60 |
+
"beta_slow": 1.0,
|
| 61 |
+
"factor": 40.0,
|
| 62 |
+
"mscale": 0.707,
|
| 63 |
+
"mscale_all_dim": 0.707,
|
| 64 |
+
"original_max_position_embeddings": 4096,
|
| 65 |
+
"rope_type": "yarn",
|
| 66 |
+
"type": "yarn"
|
| 67 |
+
},
|
| 68 |
+
"rope_theta": 10000,
|
| 69 |
+
"routed_scaling_factor": 1.0,
|
| 70 |
+
"scoring_func": "softmax",
|
| 71 |
+
"seq_aux": true,
|
| 72 |
+
"topk_group": 1,
|
| 73 |
+
"topk_method": "greedy",
|
| 74 |
+
"use_cache": true,
|
| 75 |
+
"v_head_dim": 128,
|
| 76 |
+
"vocab_size": 102400
|
| 77 |
+
},
|
| 78 |
+
"tie_word_embeddings": false,
|
| 79 |
+
"transformers_version": "4.57.3"
|
| 80 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 100000,
|
| 4 |
+
"eos_token_id": 100001,
|
| 5 |
+
"pad_token_id": 128004,
|
| 6 |
+
"transformers_version": "4.57.3"
|
| 7 |
+
}
|
model-00001-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2931cafa1fa8a14892058b7c04e215d3080aadd8deac313d65bac831fcda0722
|
| 3 |
+
size 4994768288
|
model-00002-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3e679e181f8ac20c9f2a25eb8f6099806694c4bb22e5da3811ac22c976999213
|
| 3 |
+
size 4997663496
|
model-00003-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66ca300e22398d6bf077914bb21413514e777152c897e15c3ad1062a79bb1e2a
|
| 3 |
+
size 4996097360
|
model-00004-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19016241766c0b04f7dcc380b01efd1509e53d7eb5e8c659bd32d5d69144bc5f
|
| 3 |
+
size 4990600768
|
model-00005-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c691e4a37f7ec6714529d217a08b183f6cc5c65110d7eaf13e4387d8c19063ce
|
| 3 |
+
size 4994788480
|
model-00006-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:558b8c79c7379e6b171954be3fede57746c4264270bb63e7607b8d2c84a42f55
|
| 3 |
+
size 4996098224
|
model-00007-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e876222857702d08866b10670ff27f9a88b37a11f823795b52e0b2447a0c8865
|
| 3 |
+
size 4996098224
|
model-00008-of-00008.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a0511e1db0e596ba6434614393cd4aebc161658865bda1972e9e72560fdec85
|
| 3 |
+
size 3885914632
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|