Hugging Face's logo

mlx-community
/
Hunyuan-7B-Instruct-4bit

Text Generation
Transformers
Safetensors
MLX
English
hunyuan
conversational
custom_code
4-bit precision
Model card Files Community
1
awni commited on
Commit
792efa9
·
verified ·
1 Parent(s): bb5d580

Add files using upload-large-folder tool

Browse files
Files changed (10) hide show
  1. README.md +40 -0
  2. config.json +172 -0
  3. configuration_hunyuan.py +243 -0
  4. hunyuan.tiktoken +0 -0
  5. model.safetensors +3 -0
  6. model.safetensors.index.json +811 -0
  7. modeling_hunyuan.py +0 -0
  8. special_tokens_map.json +9 -0
  9. tokenization_hy.py +296 -0
  10. tokenizer_config.json +21 -0
README.md ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ pipeline_tag: text-generation
5
+ library_name: transformers
6
+ license: other
7
+ license_name: tencent-license
8
+ license_link: https://huggingface.co/tencent/Hunyuan-7B-Instruct/blob/main/LICENSE.txt
9
+ base_model: tencent/Hunyuan-7B-Instruct
10
+ tags:
11
+ - mlx
12
+ ---
13
+
14
+ # mlx-community/Hunyuan-7B-Instruct-4bit
15
+
16
+ The Model [mlx-community/Hunyuan-7B-Instruct-4bit](https://huggingface.co/mlx-community/Hunyuan-7B-Instruct-4bit) was
17
+ converted to MLX format from [tencent/Hunyuan-7B-Instruct](https://huggingface.co/tencent/Hunyuan-7B-Instruct)
18
+ using mlx-lm version **0.21.0**.
19
+
20
+ ## Use with mlx
21
+
22
+ ```bash
23
+ pip install mlx-lm
24
+ ```
25
+
26
+ ```python
27
+ from mlx_lm import load, generate
28
+
29
+ model, tokenizer = load("mlx-community/Hunyuan-7B-Instruct-4bit")
30
+
31
+ prompt = "hello"
32
+
33
+ if tokenizer.chat_template is not None:
34
+ messages = [{"role": "user", "content": prompt}]
35
+ prompt = tokenizer.apply_chat_template(
36
+ messages, add_generation_prompt=True
37
+ )
38
+
39
+ response = generate(model, tokenizer, prompt=prompt, verbose=True)
40
+ ```
config.json ADDED
@@ -0,0 +1,172 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "HunYuanForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "attention_head_dim": 128,
8
+ "auto_map": {
9
+ "AutoConfig": "configuration_hunyuan.HunYuanConfig",
10
+ "AutoModel": "modeling_hunyuan.HunyuanModel",
11
+ "AutoModelForCausalLM": "modeling_hunyuan.HunYuanForCausalLM"
12
+ },
13
+ "bos_token_id": 1,
14
+ "cla_share_factor": 2,
15
+ "eos_token_id": 2,
16
+ "group_limited_greedy": false,
17
+ "hidden_act": "silu",
18
+ "hidden_size": 4096,
19
+ "initializer_range": 0.02,
20
+ "intermediate_size": 14336,
21
+ "kv_lora_rank": null,
22
+ "max_position_embeddings": 4096,
23
+ "mlp_bias": false,
24
+ "model_type": "hunyuan",
25
+ "moe_drop_tokens": false,
26
+ "moe_intermediate_size": [
27
+ 14336,
28
+ 14336,
29
+ 14336,
30
+ 14336,
31
+ 14336,
32
+ 14336,
33
+ 14336,
34
+ 14336,
35
+ 14336,
36
+ 14336,
37
+ 14336,
38
+ 14336,
39
+ 14336,
40
+ 14336,
41
+ 14336,
42
+ 14336,
43
+ 14336,
44
+ 14336,
45
+ 14336,
46
+ 14336,
47
+ 14336,
48
+ 14336,
49
+ 14336,
50
+ 14336,
51
+ 14336,
52
+ 14336,
53
+ 14336,
54
+ 14336,
55
+ 14336,
56
+ 14336,
57
+ 14336,
58
+ 14336
59
+ ],
60
+ "moe_layer_num_skipped": 0,
61
+ "moe_random_routing_dropped_token": false,
62
+ "moe_topk": [
63
+ 1,
64
+ 1,
65
+ 1,
66
+ 1,
67
+ 1,
68
+ 1,
69
+ 1,
70
+ 1,
71
+ 1,
72
+ 1,
73
+ 1,
74
+ 1,
75
+ 1,
76
+ 1,
77
+ 1,
78
+ 1,
79
+ 1,
80
+ 1,
81
+ 1,
82
+ 1,
83
+ 1,
84
+ 1,
85
+ 1,
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 1,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 1,
94
+ 1
95
+ ],
96
+ "n_group": false,
97
+ "norm_topk_prob": false,
98
+ "num_attention_heads": 32,
99
+ "num_experts": 1,
100
+ "num_hidden_layers": 32,
101
+ "num_key_value_heads": 8,
102
+ "num_shared_expert": [
103
+ 1,
104
+ 1,
105
+ 1,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 1,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 1,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 1,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 1,
122
+ 1,
123
+ 1,
124
+ 1,
125
+ 1,
126
+ 1,
127
+ 1,
128
+ 1,
129
+ 1,
130
+ 1,
131
+ 1,
132
+ 1,
133
+ 1,
134
+ 1
135
+ ],
136
+ "pad_token_id": 0,
137
+ "pretraining_tp": 1,
138
+ "q_lora_rank": null,
139
+ "qk_nope_head_dim": null,
140
+ "qk_rope_head_dim": null,
141
+ "quantization": {
142
+ "group_size": 64,
143
+ "bits": 4
144
+ },
145
+ "quantization_config": {
146
+ "group_size": 64,
147
+ "bits": 4
148
+ },
149
+ "rms_norm_eps": 1e-05,
150
+ "rope_scaling": {
151
+ "alpha": 1000.0,
152
+ "beta_fast": 32,
153
+ "beta_slow": 1,
154
+ "factor": 1.0,
155
+ "mscale": 1.0,
156
+ "mscale_all_dim": 1.0,
157
+ "type": "dynamic"
158
+ },
159
+ "rope_theta": 10000.0,
160
+ "routed_scaling_factor": false,
161
+ "tie_word_embeddings": true,
162
+ "topk_group": false,
163
+ "torch_dtype": "bfloat16",
164
+ "transformers_version": "4.41.2",
165
+ "use_cache": true,
166
+ "use_cla": false,
167
+ "use_mixed_mlp_moe": false,
168
+ "use_mla": false,
169
+ "use_qk_norm": true,
170
+ "v_head_dim": null,
171
+ "vocab_size": 129024
172
+ }
configuration_hunyuan.py ADDED
@@ -0,0 +1,243 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright (C) 2024 THL A29 Limited, a Tencent company. All rights reserved.
3
+ """ HunYuan model configuration"""
4
+
5
+ from transformers.configuration_utils import PretrainedConfig
6
+ from transformers.utils import logging
7
+ from typing import List, Union, Optional
8
+
9
+
10
+ logger = logging.get_logger(__name__)
11
+
12
+
13
+ class HunYuanConfig(PretrainedConfig):
14
+ r"""
15
+ This is the configuration class to store the configuration of a [`HunYuanModel`]. It is used to instantiate an
16
+ HunYuan model according to the specified arguments, defining the model architecture. Instantiating a configuration
17
+ with the defaults will yield a similar configuration to that of the HunYuan-7B.
18
+
19
+ Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
20
+ documentation from [`PretrainedConfig`] for more information.
21
+
22
+
23
+ Args:
24
+ vocab_size (`int`, *optional*, defaults to 32000):
25
+ Vocabulary size of the HunYuan model. Defines the number of different tokens that can be represented by the
26
+ `inputs_ids` passed when calling [`HunYuanModel`]
27
+ hidden_size (`int`, *optional*, defaults to 4096):
28
+ Dimension of the hidden representations.
29
+ intermediate_size (`int`, *optional*, defaults to 11008):
30
+ Dimension of the MLP representations or shared MLP representations.
31
+ moe_intermediate_size (`int` or `List`, *optional*, defaults to 11008):
32
+ Dimension of the MLP representations in MoE. Use a list if you want a different size per layer.
33
+ num_hidden_layers (`int`, *optional*, defaults to 32):
34
+ Number of hidden layers in the Transformer decoder.
35
+ num_attention_heads (`int`, *optional*, defaults to 32):
36
+ Number of attention heads for each attention layer in the Transformer decoder.
37
+ num_key_value_heads (`int`, *optional*):
38
+ This is the number of key_value heads that should be used to implement Grouped Query Attention. If
39
+ `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
40
+ `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
41
+ converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
42
+ by meanpooling all the original heads within that group. For more details checkout [this
43
+ paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
44
+ `num_attention_heads`.
45
+ hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
46
+ The non-linear activation function (function or string) in the decoder.
47
+ max_position_embeddings (`int`, *optional*, defaults to 2048):
48
+ The maximum sequence length that this model might ever be used with.
49
+ initializer_range (`float`, *optional*, defaults to 0.02):
50
+ The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
51
+ rms_norm_eps (`float`, *optional*, defaults to 1e-06):
52
+ The epsilon used by the rms normalization layers.
53
+ use_cache (`bool`, *optional*, defaults to `True`):
54
+ Whether or not the model should return the last key/values attentions (not used by all models). Only
55
+ relevant if `config.is_decoder=True`.
56
+ pad_token_id (`int`, *optional*):
57
+ Padding token id.
58
+ bos_token_id (`int`, *optional*, defaults to 1):
59
+ Beginning of stream token id.
60
+ eos_token_id (`int`, *optional*, defaults to 2):
61
+ End of stream token id.
62
+ pretraining_tp (`int`, *optional*, defaults to 1):
63
+ Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
64
+ document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
65
+ necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
66
+ issue](https://github.com/pytorch/pytorch/issues/76232).
67
+ tie_word_embeddings (`bool`, *optional*, defaults to `False`):
68
+ Whether to tie weight embeddings
69
+ rope_theta (`float`, *optional*, defaults to 10000.0):
70
+ The base period of the RoPE embeddings.
71
+ rope_scaling (`Dict`, *optional*):
72
+ Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
73
+ strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
74
+ `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
75
+ `max_position_embeddings` to the expected new maximum. See the following thread for more information on how
76
+ these scaling strategies behave:
77
+ https://www.reddit.com/r/LocalLLaMA/comments/14mrgpr/dynamically_scaled_rope_further_increases/. This is an
78
+ experimental feature, subject to breaking API changes in future versions.
79
+ attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
80
+ Whether to use a bias in the query, key, value and output projection layers during self-attention.
81
+ attention_dropout (`float`, *optional*, defaults to 0.0):
82
+ The dropout ratio for the attention probabilities.
83
+ use_qk_norm (`bool`, *optional*, defaults to `False`):
84
+ Whether query and key in attention use norm
85
+ use_cla (`bool`, *optional*, defaults to `False`):
86
+ Whether to use CLA in attention
87
+ cla_share_factor (`int`, *optional*, defaults to 1):
88
+ The share factor of CLA
89
+ num_experts (`int` or `List`, *optional*, defaults to 1):
90
+ The number of experts for moe. If it is a list, it will be used as the number of experts for each layer.
91
+ num_shared_expert (`int` or `List`, *optional*, defaults to 1):
92
+ The number of shared experts for moe. If it is a list, it will be used as the number of shared experts for each layer.
93
+ moe_topk (`int` or `List`, *optional*, defaults to 1):
94
+ The topk value for moe. If it is a list, it will be used as the topk value for each layer.
95
+ capacity_factor (Not used) (`float` or `List`, *optional*, defaults to 1.0):
96
+ The capacity factor for moe. If it is a list, it will be used as the capacity factor for each layer.
97
+ moe_layer_num_skipped (`int`, *optional*, defaults to 0):
98
+ First moe_layer_num_skipped layers do not use MoE.
99
+ """
100
+
101
+ model_type = "hunyuan"
102
+ keys_to_ignore_at_inference = ["past_key_values"]
103
+
104
+ def __init__(
105
+ self,
106
+ vocab_size=290943,
107
+ hidden_size=4096,
108
+ intermediate_size: int=11008,
109
+ moe_intermediate_size: Union[int, List]=None,
110
+ num_hidden_layers=32,
111
+ num_attention_heads=32,
112
+ num_key_value_heads=None,
113
+ attention_head_dim=None,
114
+ hidden_act="silu",
115
+ max_position_embeddings=2048,
116
+ initializer_range=0.02,
117
+ rms_norm_eps=1e-5,
118
+ use_cache=True,
119
+ pad_token_id=0,
120
+ bos_token_id=1,
121
+ eos_token_id=2,
122
+ pretraining_tp=1,
123
+ tie_word_embeddings=False,
124
+ rope_theta=10000.0,
125
+ rope_scaling=None,
126
+ attention_bias=False,
127
+ mlp_bias=False,
128
+ attention_dropout=0.0,
129
+ use_qk_norm=False,
130
+ use_cla=False,
131
+ cla_share_factor=1,
132
+ num_experts: Union[int, List]=1,
133
+ use_mixed_mlp_moe=False,
134
+ num_shared_expert: Union[int, List]=1,
135
+ moe_topk: Union[int, List]=1,
136
+ # capacity_factor: Union[int, List]=1.0,
137
+ moe_drop_tokens=False,
138
+ moe_random_routing_dropped_token=False,
139
+ use_mla=False,
140
+ kv_lora_rank=512,
141
+ q_lora_rank=1536,
142
+ qk_rope_head_dim=64,
143
+ v_head_dim=128,
144
+ qk_nope_head_dim=128,
145
+ moe_layer_num_skipped=0,
146
+ norm_topk_prob=False,
147
+ routed_scaling_factor=1.0,
148
+ group_limited_greedy=False,
149
+ n_group=None,
150
+ topk_group=None,
151
+ **kwargs,
152
+ ):
153
+ self.vocab_size = vocab_size
154
+ self.max_position_embeddings = max_position_embeddings
155
+ self.hidden_size = hidden_size
156
+ self.intermediate_size = intermediate_size
157
+ self.moe_intermediate_size = moe_intermediate_size
158
+ self.num_hidden_layers = num_hidden_layers
159
+ self.num_attention_heads = num_attention_heads
160
+ self.num_experts = num_experts
161
+ self.use_mixed_mlp_moe = use_mixed_mlp_moe
162
+ self.num_shared_expert = num_shared_expert
163
+ self.moe_topk = moe_topk
164
+ # self.capacity_factor = capacity_factor
165
+ self.moe_drop_tokens = moe_drop_tokens
166
+ self.moe_random_routing_dropped_token = moe_random_routing_dropped_token
167
+
168
+ if attention_head_dim is not None:
169
+ self.attention_head_dim = attention_head_dim
170
+ else:
171
+ self.attention_head_dim = self.hidden_size // num_attention_heads
172
+
173
+ # for backward compatibility
174
+ if num_key_value_heads is None:
175
+ num_key_value_heads = num_attention_heads
176
+
177
+ self.num_key_value_heads = num_key_value_heads
178
+ self.hidden_act = hidden_act
179
+ self.initializer_range = initializer_range
180
+ self.rms_norm_eps = rms_norm_eps
181
+ self.pretraining_tp = pretraining_tp
182
+ self.use_cache = use_cache
183
+ self.rope_theta = rope_theta
184
+ self.rope_scaling = rope_scaling
185
+ # self._rope_scaling_validation() # TODO: Need validation?
186
+ self.attention_bias = attention_bias
187
+ self.mlp_bias = mlp_bias
188
+ self.attention_dropout = attention_dropout
189
+ self.use_qk_norm = use_qk_norm
190
+ self.use_cla = use_cla
191
+ self.cla_share_factor = cla_share_factor
192
+
193
+ # MLA args
194
+ self.use_mla = use_mla
195
+ self.kv_lora_rank = kv_lora_rank
196
+ self.q_lora_rank = q_lora_rank
197
+ self.qk_rope_head_dim = qk_rope_head_dim
198
+ self.qk_nope_head_dim = qk_nope_head_dim
199
+ self.v_head_dim = v_head_dim
200
+
201
+ # DeepSeek related args
202
+ self.moe_layer_num_skipped = moe_layer_num_skipped
203
+ self.norm_topk_prob = norm_topk_prob
204
+ self.routed_scaling_factor = routed_scaling_factor
205
+ self.group_limited_greedy = group_limited_greedy
206
+ self.n_group = n_group
207
+ self.topk_group = topk_group
208
+
209
+ super().__init__(
210
+ pad_token_id=pad_token_id,
211
+ bos_token_id=bos_token_id,
212
+ eos_token_id=eos_token_id,
213
+ tie_word_embeddings=tie_word_embeddings,
214
+ **kwargs,
215
+ )
216
+
217
+ def _rope_scaling_validation(self):
218
+ """
219
+ Validate the `rope_scaling` configuration.
220
+ """
221
+ if self.rope_scaling is None:
222
+ return
223
+
224
+ if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
225
+ raise ValueError(
226
+ "`rope_scaling` must be a dictionary with with two fields, `type` and `factor` or `type` and `alpha`, "
227
+ f"got {self.rope_scaling}"
228
+ )
229
+ rope_scaling_type = self.rope_scaling.get("type", None)
230
+ rope_scaling_factor = self.rope_scaling.get("factor", None)
231
+ rope_scaling_alpha = self.rope_scaling.get("alpha", None)
232
+ if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]:
233
+ raise ValueError(
234
+ f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
235
+ )
236
+ if rope_scaling_factor is None and rope_scaling_alpha is None:
237
+ raise ValueError("`rope_scaling`'s factor or alpha field must be have one, got both of none")
238
+ if rope_scaling_factor is not None:
239
+ if not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
240
+ raise ValueError(f"`rope_scaling`'s factor field must be a float > 1.0, got {rope_scaling_factor}")
241
+ if rope_scaling_alpha is not None:
242
+ if not isinstance(rope_scaling_alpha, float) or rope_scaling_alpha <= 1.0:
243
+ raise ValueError(f"`rope_scaling`'s alpha field must be a float > 1.0, got {rope_scaling_alpha}")
hunyuan.tiktoken ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9d1aee9f7d91f05068cbed6dcca229cebada5a24dbd3c6b7f949fdf55619ddd
3
+ size 4223780188
model.safetensors.index.json ADDED
@@ -0,0 +1,811 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 4223688704
4
+ },
5
+ "weight_map": {
6
+ "model.embed_tokens.biases": "model.safetensors",
7
+ "model.embed_tokens.scales": "model.safetensors",
8
+ "model.embed_tokens.weight": "model.safetensors",
9
+ "model.layers.0.input_layernorm.weight": "model.safetensors",
10
+ "model.layers.0.mlp.down_proj.biases": "model.safetensors",
11
+ "model.layers.0.mlp.down_proj.scales": "model.safetensors",
12
+ "model.layers.0.mlp.down_proj.weight": "model.safetensors",
13
+ "model.layers.0.mlp.gate_proj.biases": "model.safetensors",
14
+ "model.layers.0.mlp.gate_proj.scales": "model.safetensors",
15
+ "model.layers.0.mlp.gate_proj.weight": "model.safetensors",
16
+ "model.layers.0.mlp.up_proj.biases": "model.safetensors",
17
+ "model.layers.0.mlp.up_proj.scales": "model.safetensors",
18
+ "model.layers.0.mlp.up_proj.weight": "model.safetensors",
19
+ "model.layers.0.post_attention_layernorm.weight": "model.safetensors",
20
+ "model.layers.0.self_attn.k_proj.biases": "model.safetensors",
21
+ "model.layers.0.self_attn.k_proj.scales": "model.safetensors",
22
+ "model.layers.0.self_attn.k_proj.weight": "model.safetensors",
23
+ "model.layers.0.self_attn.key_layernorm.weight": "model.safetensors",
24
+ "model.layers.0.self_attn.o_proj.biases": "model.safetensors",
25
+ "model.layers.0.self_attn.o_proj.scales": "model.safetensors",
26
+ "model.layers.0.self_attn.o_proj.weight": "model.safetensors",
27
+ "model.layers.0.self_attn.q_proj.biases": "model.safetensors",
28
+ "model.layers.0.self_attn.q_proj.scales": "model.safetensors",
29
+ "model.layers.0.self_attn.q_proj.weight": "model.safetensors",
30
+ "model.layers.0.self_attn.query_layernorm.weight": "model.safetensors",
31
+ "model.layers.0.self_attn.v_proj.biases": "model.safetensors",
32
+ "model.layers.0.self_attn.v_proj.scales": "model.safetensors",
33
+ "model.layers.0.self_attn.v_proj.weight": "model.safetensors",
34
+ "model.layers.1.input_layernorm.weight": "model.safetensors",
35
+ "model.layers.1.mlp.down_proj.biases": "model.safetensors",
36
+ "model.layers.1.mlp.down_proj.scales": "model.safetensors",
37
+ "model.layers.1.mlp.down_proj.weight": "model.safetensors",
38
+ "model.layers.1.mlp.gate_proj.biases": "model.safetensors",
39
+ "model.layers.1.mlp.gate_proj.scales": "model.safetensors",
40
+ "model.layers.1.mlp.gate_proj.weight": "model.safetensors",
41
+ "model.layers.1.mlp.up_proj.biases": "model.safetensors",
42
+ "model.layers.1.mlp.up_proj.scales": "model.safetensors",
43
+ "model.layers.1.mlp.up_proj.weight": "model.safetensors",
44
+ "model.layers.1.post_attention_layernorm.weight": "model.safetensors",
45
+ "model.layers.1.self_attn.k_proj.biases": "model.safetensors",
46
+ "model.layers.1.self_attn.k_proj.scales": "model.safetensors",
47
+ "model.layers.1.self_attn.k_proj.weight": "model.safetensors",
48
+ "model.layers.1.self_attn.key_layernorm.weight": "model.safetensors",
49
+ "model.layers.1.self_attn.o_proj.biases": "model.safetensors",
50
+ "model.layers.1.self_attn.o_proj.scales": "model.safetensors",
51
+ "model.layers.1.self_attn.o_proj.weight": "model.safetensors",
52
+ "model.layers.1.self_attn.q_proj.biases": "model.safetensors",
53
+ "model.layers.1.self_attn.q_proj.scales": "model.safetensors",
54
+ "model.layers.1.self_attn.q_proj.weight": "model.safetensors",
55
+ "model.layers.1.self_attn.query_layernorm.weight": "model.safetensors",
56
+ "model.layers.1.self_attn.v_proj.biases": "model.safetensors",
57
+ "model.layers.1.self_attn.v_proj.scales": "model.safetensors",
58
+ "model.layers.1.self_attn.v_proj.weight": "model.safetensors",
59
+ "model.layers.10.input_layernorm.weight": "model.safetensors",
60
+ "model.layers.10.mlp.down_proj.biases": "model.safetensors",
61
+ "model.layers.10.mlp.down_proj.scales": "model.safetensors",
62
+ "model.layers.10.mlp.down_proj.weight": "model.safetensors",
63
+ "model.layers.10.mlp.gate_proj.biases": "model.safetensors",
64
+ "model.layers.10.mlp.gate_proj.scales": "model.safetensors",
65
+ "model.layers.10.mlp.gate_proj.weight": "model.safetensors",
66
+ "model.layers.10.mlp.up_proj.biases": "model.safetensors",
67
+ "model.layers.10.mlp.up_proj.scales": "model.safetensors",
68
+ "model.layers.10.mlp.up_proj.weight": "model.safetensors",
69
+ "model.layers.10.post_attention_layernorm.weight": "model.safetensors",
70
+ "model.layers.10.self_attn.k_proj.biases": "model.safetensors",
71
+ "model.layers.10.self_attn.k_proj.scales": "model.safetensors",
72
+ "model.layers.10.self_attn.k_proj.weight": "model.safetensors",
73
+ "model.layers.10.self_attn.key_layernorm.weight": "model.safetensors",
74
+ "model.layers.10.self_attn.o_proj.biases": "model.safetensors",
75
+ "model.layers.10.self_attn.o_proj.scales": "model.safetensors",
76
+ "model.layers.10.self_attn.o_proj.weight": "model.safetensors",
77
+ "model.layers.10.self_attn.q_proj.biases": "model.safetensors",
78
+ "model.layers.10.self_attn.q_proj.scales": "model.safetensors",
79
+ "model.layers.10.self_attn.q_proj.weight": "model.safetensors",
80
+ "model.layers.10.self_attn.query_layernorm.weight": "model.safetensors",
81
+ "model.layers.10.self_attn.v_proj.biases": "model.safetensors",
82
+ "model.layers.10.self_attn.v_proj.scales": "model.safetensors",
83
+ "model.layers.10.self_attn.v_proj.weight": "model.safetensors",
84
+ "model.layers.11.input_layernorm.weight": "model.safetensors",
85
+ "model.layers.11.mlp.down_proj.biases": "model.safetensors",
86
+ "model.layers.11.mlp.down_proj.scales": "model.safetensors",
87
+ "model.layers.11.mlp.down_proj.weight": "model.safetensors",
88
+ "model.layers.11.mlp.gate_proj.biases": "model.safetensors",
89
+ "model.layers.11.mlp.gate_proj.scales": "model.safetensors",
90
+ "model.layers.11.mlp.gate_proj.weight": "model.safetensors",
91
+ "model.layers.11.mlp.up_proj.biases": "model.safetensors",
92
+ "model.layers.11.mlp.up_proj.scales": "model.safetensors",
93
+ "model.layers.11.mlp.up_proj.weight": "model.safetensors",
94
+ "model.layers.11.post_attention_layernorm.weight": "model.safetensors",
95
+ "model.layers.11.self_attn.k_proj.biases": "model.safetensors",
96
+ "model.layers.11.self_attn.k_proj.scales": "model.safetensors",
97
+ "model.layers.11.self_attn.k_proj.weight": "model.safetensors",
98
+ "model.layers.11.self_attn.key_layernorm.weight": "model.safetensors",
99
+ "model.layers.11.self_attn.o_proj.biases": "model.safetensors",
100
+ "model.layers.11.self_attn.o_proj.scales": "model.safetensors",
101
+ "model.layers.11.self_attn.o_proj.weight": "model.safetensors",
102
+ "model.layers.11.self_attn.q_proj.biases": "model.safetensors",
103
+ "model.layers.11.self_attn.q_proj.scales": "model.safetensors",
104
+ "model.layers.11.self_attn.q_proj.weight": "model.safetensors",
105
+ "model.layers.11.self_attn.query_layernorm.weight": "model.safetensors",
106
+ "model.layers.11.self_attn.v_proj.biases": "model.safetensors",
107
+ "model.layers.11.self_attn.v_proj.scales": "model.safetensors",
108
+ "model.layers.11.self_attn.v_proj.weight": "model.safetensors",
109
+ "model.layers.12.input_layernorm.weight": "model.safetensors",
110
+ "model.layers.12.mlp.down_proj.biases": "model.safetensors",
111
+ "model.layers.12.mlp.down_proj.scales": "model.safetensors",
112
+ "model.layers.12.mlp.down_proj.weight": "model.safetensors",
113
+ "model.layers.12.mlp.gate_proj.biases": "model.safetensors",
114
+ "model.layers.12.mlp.gate_proj.scales": "model.safetensors",
115
+ "model.layers.12.mlp.gate_proj.weight": "model.safetensors",
116
+ "model.layers.12.mlp.up_proj.biases": "model.safetensors",
117
+ "model.layers.12.mlp.up_proj.scales": "model.safetensors",
118
+ "model.layers.12.mlp.up_proj.weight": "model.safetensors",
119
+ "model.layers.12.post_attention_layernorm.weight": "model.safetensors",
120
+ "model.layers.12.self_attn.k_proj.biases": "model.safetensors",
121
+ "model.layers.12.self_attn.k_proj.scales": "model.safetensors",
122
+ "model.layers.12.self_attn.k_proj.weight": "model.safetensors",
123
+ "model.layers.12.self_attn.key_layernorm.weight": "model.safetensors",
124
+ "model.layers.12.self_attn.o_proj.biases": "model.safetensors",
125
+ "model.layers.12.self_attn.o_proj.scales": "model.safetensors",
126
+ "model.layers.12.self_attn.o_proj.weight": "model.safetensors",
127
+ "model.layers.12.self_attn.q_proj.biases": "model.safetensors",
128
+ "model.layers.12.self_attn.q_proj.scales": "model.safetensors",
129
+ "model.layers.12.self_attn.q_proj.weight": "model.safetensors",
130
+ "model.layers.12.self_attn.query_layernorm.weight": "model.safetensors",
131
+ "model.layers.12.self_attn.v_proj.biases": "model.safetensors",
132
+ "model.layers.12.self_attn.v_proj.scales": "model.safetensors",
133
+ "model.layers.12.self_attn.v_proj.weight": "model.safetensors",
134
+ "model.layers.13.input_layernorm.weight": "model.safetensors",
135
+ "model.layers.13.mlp.down_proj.biases": "model.safetensors",
136
+ "model.layers.13.mlp.down_proj.scales": "model.safetensors",
137
+ "model.layers.13.mlp.down_proj.weight": "model.safetensors",
138
+ "model.layers.13.mlp.gate_proj.biases": "model.safetensors",
139
+ "model.layers.13.mlp.gate_proj.scales": "model.safetensors",
140
+ "model.layers.13.mlp.gate_proj.weight": "model.safetensors",
141
+ "model.layers.13.mlp.up_proj.biases": "model.safetensors",
142
+ "model.layers.13.mlp.up_proj.scales": "model.safetensors",
143
+ "model.layers.13.mlp.up_proj.weight": "model.safetensors",
144
+ "model.layers.13.post_attention_layernorm.weight": "model.safetensors",
145
+ "model.layers.13.self_attn.k_proj.biases": "model.safetensors",
146
+ "model.layers.13.self_attn.k_proj.scales": "model.safetensors",
147
+ "model.layers.13.self_attn.k_proj.weight": "model.safetensors",
148
+ "model.layers.13.self_attn.key_layernorm.weight": "model.safetensors",
149
+ "model.layers.13.self_attn.o_proj.biases": "model.safetensors",
150
+ "model.layers.13.self_attn.o_proj.scales": "model.safetensors",
151
+ "model.layers.13.self_attn.o_proj.weight": "model.safetensors",
152
+ "model.layers.13.self_attn.q_proj.biases": "model.safetensors",
153
+ "model.layers.13.self_attn.q_proj.scales": "model.safetensors",
154
+ "model.layers.13.self_attn.q_proj.weight": "model.safetensors",
155
+ "model.layers.13.self_attn.query_layernorm.weight": "model.safetensors",
156
+ "model.layers.13.self_attn.v_proj.biases": "model.safetensors",
157
+ "model.layers.13.self_attn.v_proj.scales": "model.safetensors",
158
+ "model.layers.13.self_attn.v_proj.weight": "model.safetensors",
159
+ "model.layers.14.input_layernorm.weight": "model.safetensors",
160
+ "model.layers.14.mlp.down_proj.biases": "model.safetensors",
161
+ "model.layers.14.mlp.down_proj.scales": "model.safetensors",
162
+ "model.layers.14.mlp.down_proj.weight": "model.safetensors",
163
+ "model.layers.14.mlp.gate_proj.biases": "model.safetensors",
164
+ "model.layers.14.mlp.gate_proj.scales": "model.safetensors",
165
+ "model.layers.14.mlp.gate_proj.weight": "model.safetensors",
166
+ "model.layers.14.mlp.up_proj.biases": "model.safetensors",
167
+ "model.layers.14.mlp.up_proj.scales": "model.safetensors",
168
+ "model.layers.14.mlp.up_proj.weight": "model.safetensors",
169
+ "model.layers.14.post_attention_layernorm.weight": "model.safetensors",
170
+ "model.layers.14.self_attn.k_proj.biases": "model.safetensors",
171
+ "model.layers.14.self_attn.k_proj.scales": "model.safetensors",
172
+ "model.layers.14.self_attn.k_proj.weight": "model.safetensors",
173
+ "model.layers.14.self_attn.key_layernorm.weight": "model.safetensors",
174
+ "model.layers.14.self_attn.o_proj.biases": "model.safetensors",
175
+ "model.layers.14.self_attn.o_proj.scales": "model.safetensors",
176
+ "model.layers.14.self_attn.o_proj.weight": "model.safetensors",
177
+ "model.layers.14.self_attn.q_proj.biases": "model.safetensors",
178
+ "model.layers.14.self_attn.q_proj.scales": "model.safetensors",
179
+ "model.layers.14.self_attn.q_proj.weight": "model.safetensors",
180
+ "model.layers.14.self_attn.query_layernorm.weight": "model.safetensors",
181
+ "model.layers.14.self_attn.v_proj.biases": "model.safetensors",
182
+ "model.layers.14.self_attn.v_proj.scales": "model.safetensors",
183
+ "model.layers.14.self_attn.v_proj.weight": "model.safetensors",
184
+ "model.layers.15.input_layernorm.weight": "model.safetensors",
185
+ "model.layers.15.mlp.down_proj.biases": "model.safetensors",
186
+ "model.layers.15.mlp.down_proj.scales": "model.safetensors",
187
+ "model.layers.15.mlp.down_proj.weight": "model.safetensors",
188
+ "model.layers.15.mlp.gate_proj.biases": "model.safetensors",
189
+ "model.layers.15.mlp.gate_proj.scales": "model.safetensors",
190
+ "model.layers.15.mlp.gate_proj.weight": "model.safetensors",
191
+ "model.layers.15.mlp.up_proj.biases": "model.safetensors",
192
+ "model.layers.15.mlp.up_proj.scales": "model.safetensors",
193
+ "model.layers.15.mlp.up_proj.weight": "model.safetensors",
194
+ "model.layers.15.post_attention_layernorm.weight": "model.safetensors",
195
+ "model.layers.15.self_attn.k_proj.biases": "model.safetensors",
196
+ "model.layers.15.self_attn.k_proj.scales": "model.safetensors",
197
+ "model.layers.15.self_attn.k_proj.weight": "model.safetensors",
198
+ "model.layers.15.self_attn.key_layernorm.weight": "model.safetensors",
199
+ "model.layers.15.self_attn.o_proj.biases": "model.safetensors",
200
+ "model.layers.15.self_attn.o_proj.scales": "model.safetensors",
201
+ "model.layers.15.self_attn.o_proj.weight": "model.safetensors",
202
+ "model.layers.15.self_attn.q_proj.biases": "model.safetensors",
203
+ "model.layers.15.self_attn.q_proj.scales": "model.safetensors",
204
+ "model.layers.15.self_attn.q_proj.weight": "model.safetensors",
205
+ "model.layers.15.self_attn.query_layernorm.weight": "model.safetensors",
206
+ "model.layers.15.self_attn.v_proj.biases": "model.safetensors",
207
+ "model.layers.15.self_attn.v_proj.scales": "model.safetensors",
208
+ "model.layers.15.self_attn.v_proj.weight": "model.safetensors",
209
+ "model.layers.16.input_layernorm.weight": "model.safetensors",
210
+ "model.layers.16.mlp.down_proj.biases": "model.safetensors",
211
+ "model.layers.16.mlp.down_proj.scales": "model.safetensors",
212
+ "model.layers.16.mlp.down_proj.weight": "model.safetensors",
213
+ "model.layers.16.mlp.gate_proj.biases": "model.safetensors",
214
+ "model.layers.16.mlp.gate_proj.scales": "model.safetensors",
215
+ "model.layers.16.mlp.gate_proj.weight": "model.safetensors",
216
+ "model.layers.16.mlp.up_proj.biases": "model.safetensors",
217
+ "model.layers.16.mlp.up_proj.scales": "model.safetensors",
218
+ "model.layers.16.mlp.up_proj.weight": "model.safetensors",
219
+ "model.layers.16.post_attention_layernorm.weight": "model.safetensors",
220
+ "model.layers.16.self_attn.k_proj.biases": "model.safetensors",
221
+ "model.layers.16.self_attn.k_proj.scales": "model.safetensors",
222
+ "model.layers.16.self_attn.k_proj.weight": "model.safetensors",
223
+ "model.layers.16.self_attn.key_layernorm.weight": "model.safetensors",
224
+ "model.layers.16.self_attn.o_proj.biases": "model.safetensors",
225
+ "model.layers.16.self_attn.o_proj.scales": "model.safetensors",
226
+ "model.layers.16.self_attn.o_proj.weight": "model.safetensors",
227
+ "model.layers.16.self_attn.q_proj.biases": "model.safetensors",
228
+ "model.layers.16.self_attn.q_proj.scales": "model.safetensors",
229
+ "model.layers.16.self_attn.q_proj.weight": "model.safetensors",
230
+ "model.layers.16.self_attn.query_layernorm.weight": "model.safetensors",
231
+ "model.layers.16.self_attn.v_proj.biases": "model.safetensors",
232
+ "model.layers.16.self_attn.v_proj.scales": "model.safetensors",
233
+ "model.layers.16.self_attn.v_proj.weight": "model.safetensors",
234
+ "model.layers.17.input_layernorm.weight": "model.safetensors",
235
+ "model.layers.17.mlp.down_proj.biases": "model.safetensors",
236
+ "model.layers.17.mlp.down_proj.scales": "model.safetensors",
237
+ "model.layers.17.mlp.down_proj.weight": "model.safetensors",
238
+ "model.layers.17.mlp.gate_proj.biases": "model.safetensors",
239
+ "model.layers.17.mlp.gate_proj.scales": "model.safetensors",
240
+ "model.layers.17.mlp.gate_proj.weight": "model.safetensors",
241
+ "model.layers.17.mlp.up_proj.biases": "model.safetensors",
242
+ "model.layers.17.mlp.up_proj.scales": "model.safetensors",
243
+ "model.layers.17.mlp.up_proj.weight": "model.safetensors",
244
+ "model.layers.17.post_attention_layernorm.weight": "model.safetensors",
245
+ "model.layers.17.self_attn.k_proj.biases": "model.safetensors",
246
+ "model.layers.17.self_attn.k_proj.scales": "model.safetensors",
247
+ "model.layers.17.self_attn.k_proj.weight": "model.safetensors",
248
+ "model.layers.17.self_attn.key_layernorm.weight": "model.safetensors",
249
+ "model.layers.17.self_attn.o_proj.biases": "model.safetensors",
250
+ "model.layers.17.self_attn.o_proj.scales": "model.safetensors",
251
+ "model.layers.17.self_attn.o_proj.weight": "model.safetensors",
252
+ "model.layers.17.self_attn.q_proj.biases": "model.safetensors",
253
+ "model.layers.17.self_attn.q_proj.scales": "model.safetensors",
254
+ "model.layers.17.self_attn.q_proj.weight": "model.safetensors",
255
+ "model.layers.17.self_attn.query_layernorm.weight": "model.safetensors",
256
+ "model.layers.17.self_attn.v_proj.biases": "model.safetensors",
257
+ "model.layers.17.self_attn.v_proj.scales": "model.safetensors",
258
+ "model.layers.17.self_attn.v_proj.weight": "model.safetensors",
259
+ "model.layers.18.input_layernorm.weight": "model.safetensors",
260
+ "model.layers.18.mlp.down_proj.biases": "model.safetensors",
261
+ "model.layers.18.mlp.down_proj.scales": "model.safetensors",
262
+ "model.layers.18.mlp.down_proj.weight": "model.safetensors",
263
+ "model.layers.18.mlp.gate_proj.biases": "model.safetensors",
264
+ "model.layers.18.mlp.gate_proj.scales": "model.safetensors",
265
+ "model.layers.18.mlp.gate_proj.weight": "model.safetensors",
266
+ "model.layers.18.mlp.up_proj.biases": "model.safetensors",
267
+ "model.layers.18.mlp.up_proj.scales": "model.safetensors",
268
+ "model.layers.18.mlp.up_proj.weight": "model.safetensors",
269
+ "model.layers.18.post_attention_layernorm.weight": "model.safetensors",
270
+ "model.layers.18.self_attn.k_proj.biases": "model.safetensors",
271
+ "model.layers.18.self_attn.k_proj.scales": "model.safetensors",
272
+ "model.layers.18.self_attn.k_proj.weight": "model.safetensors",
273
+ "model.layers.18.self_attn.key_layernorm.weight": "model.safetensors",
274
+ "model.layers.18.self_attn.o_proj.biases": "model.safetensors",
275
+ "model.layers.18.self_attn.o_proj.scales": "model.safetensors",
276
+ "model.layers.18.self_attn.o_proj.weight": "model.safetensors",
277
+ "model.layers.18.self_attn.q_proj.biases": "model.safetensors",
278
+ "model.layers.18.self_attn.q_proj.scales": "model.safetensors",
279
+ "model.layers.18.self_attn.q_proj.weight": "model.safetensors",
280
+ "model.layers.18.self_attn.query_layernorm.weight": "model.safetensors",
281
+ "model.layers.18.self_attn.v_proj.biases": "model.safetensors",
282
+ "model.layers.18.self_attn.v_proj.scales": "model.safetensors",
283
+ "model.layers.18.self_attn.v_proj.weight": "model.safetensors",
284
+ "model.layers.19.input_layernorm.weight": "model.safetensors",
285
+ "model.layers.19.mlp.down_proj.biases": "model.safetensors",
286
+ "model.layers.19.mlp.down_proj.scales": "model.safetensors",
287
+ "model.layers.19.mlp.down_proj.weight": "model.safetensors",
288
+ "model.layers.19.mlp.gate_proj.biases": "model.safetensors",
289
+ "model.layers.19.mlp.gate_proj.scales": "model.safetensors",
290
+ "model.layers.19.mlp.gate_proj.weight": "model.safetensors",
291
+ "model.layers.19.mlp.up_proj.biases": "model.safetensors",
292
+ "model.layers.19.mlp.up_proj.scales": "model.safetensors",
293
+ "model.layers.19.mlp.up_proj.weight": "model.safetensors",
294
+ "model.layers.19.post_attention_layernorm.weight": "model.safetensors",
295
+ "model.layers.19.self_attn.k_proj.biases": "model.safetensors",
296
+ "model.layers.19.self_attn.k_proj.scales": "model.safetensors",
297
+ "model.layers.19.self_attn.k_proj.weight": "model.safetensors",
298
+ "model.layers.19.self_attn.key_layernorm.weight": "model.safetensors",
299
+ "model.layers.19.self_attn.o_proj.biases": "model.safetensors",
300
+ "model.layers.19.self_attn.o_proj.scales": "model.safetensors",
301
+ "model.layers.19.self_attn.o_proj.weight": "model.safetensors",
302
+ "model.layers.19.self_attn.q_proj.biases": "model.safetensors",
303
+ "model.layers.19.self_attn.q_proj.scales": "model.safetensors",
304
+ "model.layers.19.self_attn.q_proj.weight": "model.safetensors",
305
+ "model.layers.19.self_attn.query_layernorm.weight": "model.safetensors",
306
+ "model.layers.19.self_attn.v_proj.biases": "model.safetensors",
307
+ "model.layers.19.self_attn.v_proj.scales": "model.safetensors",
308
+ "model.layers.19.self_attn.v_proj.weight": "model.safetensors",
309
+ "model.layers.2.input_layernorm.weight": "model.safetensors",
310
+ "model.layers.2.mlp.down_proj.biases": "model.safetensors",
311
+ "model.layers.2.mlp.down_proj.scales": "model.safetensors",
312
+ "model.layers.2.mlp.down_proj.weight": "model.safetensors",
313
+ "model.layers.2.mlp.gate_proj.biases": "model.safetensors",
314
+ "model.layers.2.mlp.gate_proj.scales": "model.safetensors",
315
+ "model.layers.2.mlp.gate_proj.weight": "model.safetensors",
316
+ "model.layers.2.mlp.up_proj.biases": "model.safetensors",
317
+ "model.layers.2.mlp.up_proj.scales": "model.safetensors",
318
+ "model.layers.2.mlp.up_proj.weight": "model.safetensors",
319
+ "model.layers.2.post_attention_layernorm.weight": "model.safetensors",
320
+ "model.layers.2.self_attn.k_proj.biases": "model.safetensors",
321
+ "model.layers.2.self_attn.k_proj.scales": "model.safetensors",
322
+ "model.layers.2.self_attn.k_proj.weight": "model.safetensors",
323
+ "model.layers.2.self_attn.key_layernorm.weight": "model.safetensors",
324
+ "model.layers.2.self_attn.o_proj.biases": "model.safetensors",
325
+ "model.layers.2.self_attn.o_proj.scales": "model.safetensors",
326
+ "model.layers.2.self_attn.o_proj.weight": "model.safetensors",
327
+ "model.layers.2.self_attn.q_proj.biases": "model.safetensors",
328
+ "model.layers.2.self_attn.q_proj.scales": "model.safetensors",
329
+ "model.layers.2.self_attn.q_proj.weight": "model.safetensors",
330
+ "model.layers.2.self_attn.query_layernorm.weight": "model.safetensors",
331
+ "model.layers.2.self_attn.v_proj.biases": "model.safetensors",
332
+ "model.layers.2.self_attn.v_proj.scales": "model.safetensors",
333
+ "model.layers.2.self_attn.v_proj.weight": "model.safetensors",
334
+ "model.layers.20.input_layernorm.weight": "model.safetensors",
335
+ "model.layers.20.mlp.down_proj.biases": "model.safetensors",
336
+ "model.layers.20.mlp.down_proj.scales": "model.safetensors",
337
+ "model.layers.20.mlp.down_proj.weight": "model.safetensors",
338
+ "model.layers.20.mlp.gate_proj.biases": "model.safetensors",
339
+ "model.layers.20.mlp.gate_proj.scales": "model.safetensors",
340
+ "model.layers.20.mlp.gate_proj.weight": "model.safetensors",
341
+ "model.layers.20.mlp.up_proj.biases": "model.safetensors",
342
+ "model.layers.20.mlp.up_proj.scales": "model.safetensors",
343
+ "model.layers.20.mlp.up_proj.weight": "model.safetensors",
344
+ "model.layers.20.post_attention_layernorm.weight": "model.safetensors",
345
+ "model.layers.20.self_attn.k_proj.biases": "model.safetensors",
346
+ "model.layers.20.self_attn.k_proj.scales": "model.safetensors",
347
+ "model.layers.20.self_attn.k_proj.weight": "model.safetensors",
348
+ "model.layers.20.self_attn.key_layernorm.weight": "model.safetensors",
349
+ "model.layers.20.self_attn.o_proj.biases": "model.safetensors",
350
+ "model.layers.20.self_attn.o_proj.scales": "model.safetensors",
351
+ "model.layers.20.self_attn.o_proj.weight": "model.safetensors",
352
+ "model.layers.20.self_attn.q_proj.biases": "model.safetensors",
353
+ "model.layers.20.self_attn.q_proj.scales": "model.safetensors",
354
+ "model.layers.20.self_attn.q_proj.weight": "model.safetensors",
355
+ "model.layers.20.self_attn.query_layernorm.weight": "model.safetensors",
356
+ "model.layers.20.self_attn.v_proj.biases": "model.safetensors",
357
+ "model.layers.20.self_attn.v_proj.scales": "model.safetensors",
358
+ "model.layers.20.self_attn.v_proj.weight": "model.safetensors",
359
+ "model.layers.21.input_layernorm.weight": "model.safetensors",
360
+ "model.layers.21.mlp.down_proj.biases": "model.safetensors",
361
+ "model.layers.21.mlp.down_proj.scales": "model.safetensors",
362
+ "model.layers.21.mlp.down_proj.weight": "model.safetensors",
363
+ "model.layers.21.mlp.gate_proj.biases": "model.safetensors",
364
+ "model.layers.21.mlp.gate_proj.scales": "model.safetensors",
365
+ "model.layers.21.mlp.gate_proj.weight": "model.safetensors",
366
+ "model.layers.21.mlp.up_proj.biases": "model.safetensors",
367
+ "model.layers.21.mlp.up_proj.scales": "model.safetensors",
368
+ "model.layers.21.mlp.up_proj.weight": "model.safetensors",
369
+ "model.layers.21.post_attention_layernorm.weight": "model.safetensors",
370
+ "model.layers.21.self_attn.k_proj.biases": "model.safetensors",
371
+ "model.layers.21.self_attn.k_proj.scales": "model.safetensors",
372
+ "model.layers.21.self_attn.k_proj.weight": "model.safetensors",
373
+ "model.layers.21.self_attn.key_layernorm.weight": "model.safetensors",
374
+ "model.layers.21.self_attn.o_proj.biases": "model.safetensors",
375
+ "model.layers.21.self_attn.o_proj.scales": "model.safetensors",
376
+ "model.layers.21.self_attn.o_proj.weight": "model.safetensors",
377
+ "model.layers.21.self_attn.q_proj.biases": "model.safetensors",
378
+ "model.layers.21.self_attn.q_proj.scales": "model.safetensors",
379
+ "model.layers.21.self_attn.q_proj.weight": "model.safetensors",
380
+ "model.layers.21.self_attn.query_layernorm.weight": "model.safetensors",
381
+ "model.layers.21.self_attn.v_proj.biases": "model.safetensors",
382
+ "model.layers.21.self_attn.v_proj.scales": "model.safetensors",
383
+ "model.layers.21.self_attn.v_proj.weight": "model.safetensors",
384
+ "model.layers.22.input_layernorm.weight": "model.safetensors",
385
+ "model.layers.22.mlp.down_proj.biases": "model.safetensors",
386
+ "model.layers.22.mlp.down_proj.scales": "model.safetensors",
387
+ "model.layers.22.mlp.down_proj.weight": "model.safetensors",
388
+ "model.layers.22.mlp.gate_proj.biases": "model.safetensors",
389
+ "model.layers.22.mlp.gate_proj.scales": "model.safetensors",
390
+ "model.layers.22.mlp.gate_proj.weight": "model.safetensors",
391
+ "model.layers.22.mlp.up_proj.biases": "model.safetensors",
392
+ "model.layers.22.mlp.up_proj.scales": "model.safetensors",
393
+ "model.layers.22.mlp.up_proj.weight": "model.safetensors",
394
+ "model.layers.22.post_attention_layernorm.weight": "model.safetensors",
395
+ "model.layers.22.self_attn.k_proj.biases": "model.safetensors",
396
+ "model.layers.22.self_attn.k_proj.scales": "model.safetensors",
397
+ "model.layers.22.self_attn.k_proj.weight": "model.safetensors",
398
+ "model.layers.22.self_attn.key_layernorm.weight": "model.safetensors",
399
+ "model.layers.22.self_attn.o_proj.biases": "model.safetensors",
400
+ "model.layers.22.self_attn.o_proj.scales": "model.safetensors",
401
+ "model.layers.22.self_attn.o_proj.weight": "model.safetensors",
402
+ "model.layers.22.self_attn.q_proj.biases": "model.safetensors",
403
+ "model.layers.22.self_attn.q_proj.scales": "model.safetensors",
404
+ "model.layers.22.self_attn.q_proj.weight": "model.safetensors",
405
+ "model.layers.22.self_attn.query_layernorm.weight": "model.safetensors",
406
+ "model.layers.22.self_attn.v_proj.biases": "model.safetensors",
407
+ "model.layers.22.self_attn.v_proj.scales": "model.safetensors",
408
+ "model.layers.22.self_attn.v_proj.weight": "model.safetensors",
409
+ "model.layers.23.input_layernorm.weight": "model.safetensors",
410
+ "model.layers.23.mlp.down_proj.biases": "model.safetensors",
411
+ "model.layers.23.mlp.down_proj.scales": "model.safetensors",
412
+ "model.layers.23.mlp.down_proj.weight": "model.safetensors",
413
+ "model.layers.23.mlp.gate_proj.biases": "model.safetensors",
414
+ "model.layers.23.mlp.gate_proj.scales": "model.safetensors",
415
+ "model.layers.23.mlp.gate_proj.weight": "model.safetensors",
416
+ "model.layers.23.mlp.up_proj.biases": "model.safetensors",
417
+ "model.layers.23.mlp.up_proj.scales": "model.safetensors",
418
+ "model.layers.23.mlp.up_proj.weight": "model.safetensors",
419
+ "model.layers.23.post_attention_layernorm.weight": "model.safetensors",
420
+ "model.layers.23.self_attn.k_proj.biases": "model.safetensors",
421
+ "model.layers.23.self_attn.k_proj.scales": "model.safetensors",
422
+ "model.layers.23.self_attn.k_proj.weight": "model.safetensors",
423
+ "model.layers.23.self_attn.key_layernorm.weight": "model.safetensors",
424
+ "model.layers.23.self_attn.o_proj.biases": "model.safetensors",
425
+ "model.layers.23.self_attn.o_proj.scales": "model.safetensors",
426
+ "model.layers.23.self_attn.o_proj.weight": "model.safetensors",
427
+ "model.layers.23.self_attn.q_proj.biases": "model.safetensors",
428
+ "model.layers.23.self_attn.q_proj.scales": "model.safetensors",
429
+ "model.layers.23.self_attn.q_proj.weight": "model.safetensors",
430
+ "model.layers.23.self_attn.query_layernorm.weight": "model.safetensors",
431
+ "model.layers.23.self_attn.v_proj.biases": "model.safetensors",
432
+ "model.layers.23.self_attn.v_proj.scales": "model.safetensors",
433
+ "model.layers.23.self_attn.v_proj.weight": "model.safetensors",
434
+ "model.layers.24.input_layernorm.weight": "model.safetensors",
435
+ "model.layers.24.mlp.down_proj.biases": "model.safetensors",
436
+ "model.layers.24.mlp.down_proj.scales": "model.safetensors",
437
+ "model.layers.24.mlp.down_proj.weight": "model.safetensors",
438
+ "model.layers.24.mlp.gate_proj.biases": "model.safetensors",
439
+ "model.layers.24.mlp.gate_proj.scales": "model.safetensors",
440
+ "model.layers.24.mlp.gate_proj.weight": "model.safetensors",
441
+ "model.layers.24.mlp.up_proj.biases": "model.safetensors",
442
+ "model.layers.24.mlp.up_proj.scales": "model.safetensors",
443
+ "model.layers.24.mlp.up_proj.weight": "model.safetensors",
444
+ "model.layers.24.post_attention_layernorm.weight": "model.safetensors",
445
+ "model.layers.24.self_attn.k_proj.biases": "model.safetensors",
446
+ "model.layers.24.self_attn.k_proj.scales": "model.safetensors",
447
+ "model.layers.24.self_attn.k_proj.weight": "model.safetensors",
448
+ "model.layers.24.self_attn.key_layernorm.weight": "model.safetensors",
449
+ "model.layers.24.self_attn.o_proj.biases": "model.safetensors",
450
+ "model.layers.24.self_attn.o_proj.scales": "model.safetensors",
451
+ "model.layers.24.self_attn.o_proj.weight": "model.safetensors",
452
+ "model.layers.24.self_attn.q_proj.biases": "model.safetensors",
453
+ "model.layers.24.self_attn.q_proj.scales": "model.safetensors",
454
+ "model.layers.24.self_attn.q_proj.weight": "model.safetensors",
455
+ "model.layers.24.self_attn.query_layernorm.weight": "model.safetensors",
456
+ "model.layers.24.self_attn.v_proj.biases": "model.safetensors",
457
+ "model.layers.24.self_attn.v_proj.scales": "model.safetensors",
458
+ "model.layers.24.self_attn.v_proj.weight": "model.safetensors",
459
+ "model.layers.25.input_layernorm.weight": "model.safetensors",
460
+ "model.layers.25.mlp.down_proj.biases": "model.safetensors",
461
+ "model.layers.25.mlp.down_proj.scales": "model.safetensors",
462
+ "model.layers.25.mlp.down_proj.weight": "model.safetensors",
463
+ "model.layers.25.mlp.gate_proj.biases": "model.safetensors",
464
+ "model.layers.25.mlp.gate_proj.scales": "model.safetensors",
465
+ "model.layers.25.mlp.gate_proj.weight": "model.safetensors",
466
+ "model.layers.25.mlp.up_proj.biases": "model.safetensors",
467
+ "model.layers.25.mlp.up_proj.scales": "model.safetensors",
468
+ "model.layers.25.mlp.up_proj.weight": "model.safetensors",
469
+ "model.layers.25.post_attention_layernorm.weight": "model.safetensors",
470
+ "model.layers.25.self_attn.k_proj.biases": "model.safetensors",
471
+ "model.layers.25.self_attn.k_proj.scales": "model.safetensors",
472
+ "model.layers.25.self_attn.k_proj.weight": "model.safetensors",
473
+ "model.layers.25.self_attn.key_layernorm.weight": "model.safetensors",
474
+ "model.layers.25.self_attn.o_proj.biases": "model.safetensors",
475
+ "model.layers.25.self_attn.o_proj.scales": "model.safetensors",
476
+ "model.layers.25.self_attn.o_proj.weight": "model.safetensors",
477
+ "model.layers.25.self_attn.q_proj.biases": "model.safetensors",
478
+ "model.layers.25.self_attn.q_proj.scales": "model.safetensors",
479
+ "model.layers.25.self_attn.q_proj.weight": "model.safetensors",
480
+ "model.layers.25.self_attn.query_layernorm.weight": "model.safetensors",
481
+ "model.layers.25.self_attn.v_proj.biases": "model.safetensors",
482
+ "model.layers.25.self_attn.v_proj.scales": "model.safetensors",
483
+ "model.layers.25.self_attn.v_proj.weight": "model.safetensors",
484
+ "model.layers.26.input_layernorm.weight": "model.safetensors",
485
+ "model.layers.26.mlp.down_proj.biases": "model.safetensors",
486
+ "model.layers.26.mlp.down_proj.scales": "model.safetensors",
487
+ "model.layers.26.mlp.down_proj.weight": "model.safetensors",
488
+ "model.layers.26.mlp.gate_proj.biases": "model.safetensors",
489
+ "model.layers.26.mlp.gate_proj.scales": "model.safetensors",
490
+ "model.layers.26.mlp.gate_proj.weight": "model.safetensors",
491
+ "model.layers.26.mlp.up_proj.biases": "model.safetensors",
492
+ "model.layers.26.mlp.up_proj.scales": "model.safetensors",
493
+ "model.layers.26.mlp.up_proj.weight": "model.safetensors",
494
+ "model.layers.26.post_attention_layernorm.weight": "model.safetensors",
495
+ "model.layers.26.self_attn.k_proj.biases": "model.safetensors",
496
+ "model.layers.26.self_attn.k_proj.scales": "model.safetensors",
497
+ "model.layers.26.self_attn.k_proj.weight": "model.safetensors",
498
+ "model.layers.26.self_attn.key_layernorm.weight": "model.safetensors",
499
+ "model.layers.26.self_attn.o_proj.biases": "model.safetensors",
500
+ "model.layers.26.self_attn.o_proj.scales": "model.safetensors",
501
+ "model.layers.26.self_attn.o_proj.weight": "model.safetensors",
502
+ "model.layers.26.self_attn.q_proj.biases": "model.safetensors",
503
+ "model.layers.26.self_attn.q_proj.scales": "model.safetensors",
504
+ "model.layers.26.self_attn.q_proj.weight": "model.safetensors",
505
+ "model.layers.26.self_attn.query_layernorm.weight": "model.safetensors",
506
+ "model.layers.26.self_attn.v_proj.biases": "model.safetensors",
507
+ "model.layers.26.self_attn.v_proj.scales": "model.safetensors",
508
+ "model.layers.26.self_attn.v_proj.weight": "model.safetensors",
509
+ "model.layers.27.input_layernorm.weight": "model.safetensors",
510
+ "model.layers.27.mlp.down_proj.biases": "model.safetensors",
511
+ "model.layers.27.mlp.down_proj.scales": "model.safetensors",
512
+ "model.layers.27.mlp.down_proj.weight": "model.safetensors",
513
+ "model.layers.27.mlp.gate_proj.biases": "model.safetensors",
514
+ "model.layers.27.mlp.gate_proj.scales": "model.safetensors",
515
+ "model.layers.27.mlp.gate_proj.weight": "model.safetensors",
516
+ "model.layers.27.mlp.up_proj.biases": "model.safetensors",
517
+ "model.layers.27.mlp.up_proj.scales": "model.safetensors",
518
+ "model.layers.27.mlp.up_proj.weight": "model.safetensors",
519
+ "model.layers.27.post_attention_layernorm.weight": "model.safetensors",
520
+ "model.layers.27.self_attn.k_proj.biases": "model.safetensors",
521
+ "model.layers.27.self_attn.k_proj.scales": "model.safetensors",
522
+ "model.layers.27.self_attn.k_proj.weight": "model.safetensors",
523
+ "model.layers.27.self_attn.key_layernorm.weight": "model.safetensors",
524
+ "model.layers.27.self_attn.o_proj.biases": "model.safetensors",
525
+ "model.layers.27.self_attn.o_proj.scales": "model.safetensors",
526
+ "model.layers.27.self_attn.o_proj.weight": "model.safetensors",
527
+ "model.layers.27.self_attn.q_proj.biases": "model.safetensors",
528
+ "model.layers.27.self_attn.q_proj.scales": "model.safetensors",
529
+ "model.layers.27.self_attn.q_proj.weight": "model.safetensors",
530
+ "model.layers.27.self_attn.query_layernorm.weight": "model.safetensors",
531
+ "model.layers.27.self_attn.v_proj.biases": "model.safetensors",
532
+ "model.layers.27.self_attn.v_proj.scales": "model.safetensors",
533
+ "model.layers.27.self_attn.v_proj.weight": "model.safetensors",
534
+ "model.layers.28.input_layernorm.weight": "model.safetensors",
535
+ "model.layers.28.mlp.down_proj.biases": "model.safetensors",
536
+ "model.layers.28.mlp.down_proj.scales": "model.safetensors",
537
+ "model.layers.28.mlp.down_proj.weight": "model.safetensors",
538
+ "model.layers.28.mlp.gate_proj.biases": "model.safetensors",
539
+ "model.layers.28.mlp.gate_proj.scales": "model.safetensors",
540
+ "model.layers.28.mlp.gate_proj.weight": "model.safetensors",
541
+ "model.layers.28.mlp.up_proj.biases": "model.safetensors",
542
+ "model.layers.28.mlp.up_proj.scales": "model.safetensors",
543
+ "model.layers.28.mlp.up_proj.weight": "model.safetensors",
544
+ "model.layers.28.post_attention_layernorm.weight": "model.safetensors",
545
+ "model.layers.28.self_attn.k_proj.biases": "model.safetensors",
546
+ "model.layers.28.self_attn.k_proj.scales": "model.safetensors",
547
+ "model.layers.28.self_attn.k_proj.weight": "model.safetensors",
548
+ "model.layers.28.self_attn.key_layernorm.weight": "model.safetensors",
549
+ "model.layers.28.self_attn.o_proj.biases": "model.safetensors",
550
+ "model.layers.28.self_attn.o_proj.scales": "model.safetensors",
551
+ "model.layers.28.self_attn.o_proj.weight": "model.safetensors",
552
+ "model.layers.28.self_attn.q_proj.biases": "model.safetensors",
553
+ "model.layers.28.self_attn.q_proj.scales": "model.safetensors",
554
+ "model.layers.28.self_attn.q_proj.weight": "model.safetensors",
555
+ "model.layers.28.self_attn.query_layernorm.weight": "model.safetensors",
556
+ "model.layers.28.self_attn.v_proj.biases": "model.safetensors",
557
+ "model.layers.28.self_attn.v_proj.scales": "model.safetensors",
558
+ "model.layers.28.self_attn.v_proj.weight": "model.safetensors",
559
+ "model.layers.29.input_layernorm.weight": "model.safetensors",
560
+ "model.layers.29.mlp.down_proj.biases": "model.safetensors",
561
+ "model.layers.29.mlp.down_proj.scales": "model.safetensors",
562
+ "model.layers.29.mlp.down_proj.weight": "model.safetensors",
563
+ "model.layers.29.mlp.gate_proj.biases": "model.safetensors",
564
+ "model.layers.29.mlp.gate_proj.scales": "model.safetensors",
565
+ "model.layers.29.mlp.gate_proj.weight": "model.safetensors",
566
+ "model.layers.29.mlp.up_proj.biases": "model.safetensors",
567
+ "model.layers.29.mlp.up_proj.scales": "model.safetensors",
568
+ "model.layers.29.mlp.up_proj.weight": "model.safetensors",
569
+ "model.layers.29.post_attention_layernorm.weight": "model.safetensors",
570
+ "model.layers.29.self_attn.k_proj.biases": "model.safetensors",
571
+ "model.layers.29.self_attn.k_proj.scales": "model.safetensors",
572
+ "model.layers.29.self_attn.k_proj.weight": "model.safetensors",
573
+ "model.layers.29.self_attn.key_layernorm.weight": "model.safetensors",
574
+ "model.layers.29.self_attn.o_proj.biases": "model.safetensors",
575
+ "model.layers.29.self_attn.o_proj.scales": "model.safetensors",
576
+ "model.layers.29.self_attn.o_proj.weight": "model.safetensors",
577
+ "model.layers.29.self_attn.q_proj.biases": "model.safetensors",
578
+ "model.layers.29.self_attn.q_proj.scales": "model.safetensors",
579
+ "model.layers.29.self_attn.q_proj.weight": "model.safetensors",
580
+ "model.layers.29.self_attn.query_layernorm.weight": "model.safetensors",
581
+ "model.layers.29.self_attn.v_proj.biases": "model.safetensors",
582
+ "model.layers.29.self_attn.v_proj.scales": "model.safetensors",
583
+ "model.layers.29.self_attn.v_proj.weight": "model.safetensors",
584
+ "model.layers.3.input_layernorm.weight": "model.safetensors",
585
+ "model.layers.3.mlp.down_proj.biases": "model.safetensors",
586
+ "model.layers.3.mlp.down_proj.scales": "model.safetensors",
587
+ "model.layers.3.mlp.down_proj.weight": "model.safetensors",
588
+ "model.layers.3.mlp.gate_proj.biases": "model.safetensors",
589
+ "model.layers.3.mlp.gate_proj.scales": "model.safetensors",
590
+ "model.layers.3.mlp.gate_proj.weight": "model.safetensors",
591
+ "model.layers.3.mlp.up_proj.biases": "model.safetensors",
592
+ "model.layers.3.mlp.up_proj.scales": "model.safetensors",
593
+ "model.layers.3.mlp.up_proj.weight": "model.safetensors",
594
+ "model.layers.3.post_attention_layernorm.weight": "model.safetensors",
595
+ "model.layers.3.self_attn.k_proj.biases": "model.safetensors",
596
+ "model.layers.3.self_attn.k_proj.scales": "model.safetensors",
597
+ "model.layers.3.self_attn.k_proj.weight": "model.safetensors",
598
+ "model.layers.3.self_attn.key_layernorm.weight": "model.safetensors",
599
+ "model.layers.3.self_attn.o_proj.biases": "model.safetensors",
600
+ "model.layers.3.self_attn.o_proj.scales": "model.safetensors",
601
+ "model.layers.3.self_attn.o_proj.weight": "model.safetensors",
602
+ "model.layers.3.self_attn.q_proj.biases": "model.safetensors",
603
+ "model.layers.3.self_attn.q_proj.scales": "model.safetensors",
604
+ "model.layers.3.self_attn.q_proj.weight": "model.safetensors",
605
+ "model.layers.3.self_attn.query_layernorm.weight": "model.safetensors",
606
+ "model.layers.3.self_attn.v_proj.biases": "model.safetensors",
607
+ "model.layers.3.self_attn.v_proj.scales": "model.safetensors",
608
+ "model.layers.3.self_attn.v_proj.weight": "model.safetensors",
609
+ "model.layers.30.input_layernorm.weight": "model.safetensors",
610
+ "model.layers.30.mlp.down_proj.biases": "model.safetensors",
611
+ "model.layers.30.mlp.down_proj.scales": "model.safetensors",
612
+ "model.layers.30.mlp.down_proj.weight": "model.safetensors",
613
+ "model.layers.30.mlp.gate_proj.biases": "model.safetensors",
614
+ "model.layers.30.mlp.gate_proj.scales": "model.safetensors",
615
+ "model.layers.30.mlp.gate_proj.weight": "model.safetensors",
616
+ "model.layers.30.mlp.up_proj.biases": "model.safetensors",
617
+ "model.layers.30.mlp.up_proj.scales": "model.safetensors",
618
+ "model.layers.30.mlp.up_proj.weight": "model.safetensors",
619
+ "model.layers.30.post_attention_layernorm.weight": "model.safetensors",
620
+ "model.layers.30.self_attn.k_proj.biases": "model.safetensors",
621
+ "model.layers.30.self_attn.k_proj.scales": "model.safetensors",
622
+ "model.layers.30.self_attn.k_proj.weight": "model.safetensors",
623
+ "model.layers.30.self_attn.key_layernorm.weight": "model.safetensors",
624
+ "model.layers.30.self_attn.o_proj.biases": "model.safetensors",
625
+ "model.layers.30.self_attn.o_proj.scales": "model.safetensors",
626
+ "model.layers.30.self_attn.o_proj.weight": "model.safetensors",
627
+ "model.layers.30.self_attn.q_proj.biases": "model.safetensors",
628
+ "model.layers.30.self_attn.q_proj.scales": "model.safetensors",
629
+ "model.layers.30.self_attn.q_proj.weight": "model.safetensors",
630
+ "model.layers.30.self_attn.query_layernorm.weight": "model.safetensors",
631
+ "model.layers.30.self_attn.v_proj.biases": "model.safetensors",
632
+ "model.layers.30.self_attn.v_proj.scales": "model.safetensors",
633
+ "model.layers.30.self_attn.v_proj.weight": "model.safetensors",
634
+ "model.layers.31.input_layernorm.weight": "model.safetensors",
635
+ "model.layers.31.mlp.down_proj.biases": "model.safetensors",
636
+ "model.layers.31.mlp.down_proj.scales": "model.safetensors",
637
+ "model.layers.31.mlp.down_proj.weight": "model.safetensors",
638
+ "model.layers.31.mlp.gate_proj.biases": "model.safetensors",
639
+ "model.layers.31.mlp.gate_proj.scales": "model.safetensors",
640
+ "model.layers.31.mlp.gate_proj.weight": "model.safetensors",
641
+ "model.layers.31.mlp.up_proj.biases": "model.safetensors",
642
+ "model.layers.31.mlp.up_proj.scales": "model.safetensors",
643
+ "model.layers.31.mlp.up_proj.weight": "model.safetensors",
644
+ "model.layers.31.post_attention_layernorm.weight": "model.safetensors",
645
+ "model.layers.31.self_attn.k_proj.biases": "model.safetensors",
646
+ "model.layers.31.self_attn.k_proj.scales": "model.safetensors",
647
+ "model.layers.31.self_attn.k_proj.weight": "model.safetensors",
648
+ "model.layers.31.self_attn.key_layernorm.weight": "model.safetensors",
649
+ "model.layers.31.self_attn.o_proj.biases": "model.safetensors",
650
+ "model.layers.31.self_attn.o_proj.scales": "model.safetensors",
651
+ "model.layers.31.self_attn.o_proj.weight": "model.safetensors",
652
+ "model.layers.31.self_attn.q_proj.biases": "model.safetensors",
653
+ "model.layers.31.self_attn.q_proj.scales": "model.safetensors",
654
+ "model.layers.31.self_attn.q_proj.weight": "model.safetensors",
655
+ "model.layers.31.self_attn.query_layernorm.weight": "model.safetensors",
656
+ "model.layers.31.self_attn.v_proj.biases": "model.safetensors",
657
+ "model.layers.31.self_attn.v_proj.scales": "model.safetensors",
658
+ "model.layers.31.self_attn.v_proj.weight": "model.safetensors",
659
+ "model.layers.4.input_layernorm.weight": "model.safetensors",
660
+ "model.layers.4.mlp.down_proj.biases": "model.safetensors",
661
+ "model.layers.4.mlp.down_proj.scales": "model.safetensors",
662
+ "model.layers.4.mlp.down_proj.weight": "model.safetensors",
663
+ "model.layers.4.mlp.gate_proj.biases": "model.safetensors",
664
+ "model.layers.4.mlp.gate_proj.scales": "model.safetensors",
665
+ "model.layers.4.mlp.gate_proj.weight": "model.safetensors",
666
+ "model.layers.4.mlp.up_proj.biases": "model.safetensors",
667
+ "model.layers.4.mlp.up_proj.scales": "model.safetensors",
668
+ "model.layers.4.mlp.up_proj.weight": "model.safetensors",
669
+ "model.layers.4.post_attention_layernorm.weight": "model.safetensors",
670
+ "model.layers.4.self_attn.k_proj.biases": "model.safetensors",
671
+ "model.layers.4.self_attn.k_proj.scales": "model.safetensors",
672
+ "model.layers.4.self_attn.k_proj.weight": "model.safetensors",
673
+ "model.layers.4.self_attn.key_layernorm.weight": "model.safetensors",
674
+ "model.layers.4.self_attn.o_proj.biases": "model.safetensors",
675
+ "model.layers.4.self_attn.o_proj.scales": "model.safetensors",
676
+ "model.layers.4.self_attn.o_proj.weight": "model.safetensors",
677
+ "model.layers.4.self_attn.q_proj.biases": "model.safetensors",
678
+ "model.layers.4.self_attn.q_proj.scales": "model.safetensors",
679
+ "model.layers.4.self_attn.q_proj.weight": "model.safetensors",
680
+ "model.layers.4.self_attn.query_layernorm.weight": "model.safetensors",
681
+ "model.layers.4.self_attn.v_proj.biases": "model.safetensors",
682
+ "model.layers.4.self_attn.v_proj.scales": "model.safetensors",
683
+ "model.layers.4.self_attn.v_proj.weight": "model.safetensors",
684
+ "model.layers.5.input_layernorm.weight": "model.safetensors",
685
+ "model.layers.5.mlp.down_proj.biases": "model.safetensors",
686
+ "model.layers.5.mlp.down_proj.scales": "model.safetensors",
687
+ "model.layers.5.mlp.down_proj.weight": "model.safetensors",
688
+ "model.layers.5.mlp.gate_proj.biases": "model.safetensors",
689
+ "model.layers.5.mlp.gate_proj.scales": "model.safetensors",
690
+ "model.layers.5.mlp.gate_proj.weight": "model.safetensors",
691
+ "model.layers.5.mlp.up_proj.biases": "model.safetensors",
692
+ "model.layers.5.mlp.up_proj.scales": "model.safetensors",
693
+ "model.layers.5.mlp.up_proj.weight": "model.safetensors",
694
+ "model.layers.5.post_attention_layernorm.weight": "model.safetensors",
695
+ "model.layers.5.self_attn.k_proj.biases": "model.safetensors",
696
+ "model.layers.5.self_attn.k_proj.scales": "model.safetensors",
697
+ "model.layers.5.self_attn.k_proj.weight": "model.safetensors",
698
+ "model.layers.5.self_attn.key_layernorm.weight": "model.safetensors",
699
+ "model.layers.5.self_attn.o_proj.biases": "model.safetensors",
700
+ "model.layers.5.self_attn.o_proj.scales": "model.safetensors",
701
+ "model.layers.5.self_attn.o_proj.weight": "model.safetensors",
702
+ "model.layers.5.self_attn.q_proj.biases": "model.safetensors",
703
+ "model.layers.5.self_attn.q_proj.scales": "model.safetensors",
704
+ "model.layers.5.self_attn.q_proj.weight": "model.safetensors",
705
+ "model.layers.5.self_attn.query_layernorm.weight": "model.safetensors",
706
+ "model.layers.5.self_attn.v_proj.biases": "model.safetensors",
707
+ "model.layers.5.self_attn.v_proj.scales": "model.safetensors",
708
+ "model.layers.5.self_attn.v_proj.weight": "model.safetensors",
709
+ "model.layers.6.input_layernorm.weight": "model.safetensors",
710
+ "model.layers.6.mlp.down_proj.biases": "model.safetensors",
711
+ "model.layers.6.mlp.down_proj.scales": "model.safetensors",
712
+ "model.layers.6.mlp.down_proj.weight": "model.safetensors",
713
+ "model.layers.6.mlp.gate_proj.biases": "model.safetensors",
714
+ "model.layers.6.mlp.gate_proj.scales": "model.safetensors",
715
+ "model.layers.6.mlp.gate_proj.weight": "model.safetensors",
716
+ "model.layers.6.mlp.up_proj.biases": "model.safetensors",
717
+ "model.layers.6.mlp.up_proj.scales": "model.safetensors",
718
+ "model.layers.6.mlp.up_proj.weight": "model.safetensors",
719
+ "model.layers.6.post_attention_layernorm.weight": "model.safetensors",
720
+ "model.layers.6.self_attn.k_proj.biases": "model.safetensors",
721
+ "model.layers.6.self_attn.k_proj.scales": "model.safetensors",
722
+ "model.layers.6.self_attn.k_proj.weight": "model.safetensors",
723
+ "model.layers.6.self_attn.key_layernorm.weight": "model.safetensors",
724
+ "model.layers.6.self_attn.o_proj.biases": "model.safetensors",
725
+ "model.layers.6.self_attn.o_proj.scales": "model.safetensors",
726
+ "model.layers.6.self_attn.o_proj.weight": "model.safetensors",
727
+ "model.layers.6.self_attn.q_proj.biases": "model.safetensors",
728
+ "model.layers.6.self_attn.q_proj.scales": "model.safetensors",
729
+ "model.layers.6.self_attn.q_proj.weight": "model.safetensors",
730
+ "model.layers.6.self_attn.query_layernorm.weight": "model.safetensors",
731
+ "model.layers.6.self_attn.v_proj.biases": "model.safetensors",
732
+ "model.layers.6.self_attn.v_proj.scales": "model.safetensors",
733
+ "model.layers.6.self_attn.v_proj.weight": "model.safetensors",
734
+ "model.layers.7.input_layernorm.weight": "model.safetensors",
735
+ "model.layers.7.mlp.down_proj.biases": "model.safetensors",
736
+ "model.layers.7.mlp.down_proj.scales": "model.safetensors",
737
+ "model.layers.7.mlp.down_proj.weight": "model.safetensors",
738
+ "model.layers.7.mlp.gate_proj.biases": "model.safetensors",
739
+ "model.layers.7.mlp.gate_proj.scales": "model.safetensors",
740
+ "model.layers.7.mlp.gate_proj.weight": "model.safetensors",
741
+ "model.layers.7.mlp.up_proj.biases": "model.safetensors",
742
+ "model.layers.7.mlp.up_proj.scales": "model.safetensors",
743
+ "model.layers.7.mlp.up_proj.weight": "model.safetensors",
744
+ "model.layers.7.post_attention_layernorm.weight": "model.safetensors",
745
+ "model.layers.7.self_attn.k_proj.biases": "model.safetensors",
746
+ "model.layers.7.self_attn.k_proj.scales": "model.safetensors",
747
+ "model.layers.7.self_attn.k_proj.weight": "model.safetensors",
748
+ "model.layers.7.self_attn.key_layernorm.weight": "model.safetensors",
749
+ "model.layers.7.self_attn.o_proj.biases": "model.safetensors",
750
+ "model.layers.7.self_attn.o_proj.scales": "model.safetensors",
751
+ "model.layers.7.self_attn.o_proj.weight": "model.safetensors",
752
+ "model.layers.7.self_attn.q_proj.biases": "model.safetensors",
753
+ "model.layers.7.self_attn.q_proj.scales": "model.safetensors",
754
+ "model.layers.7.self_attn.q_proj.weight": "model.safetensors",
755
+ "model.layers.7.self_attn.query_layernorm.weight": "model.safetensors",
756
+ "model.layers.7.self_attn.v_proj.biases": "model.safetensors",
757
+ "model.layers.7.self_attn.v_proj.scales": "model.safetensors",
758
+ "model.layers.7.self_attn.v_proj.weight": "model.safetensors",
759
+ "model.layers.8.input_layernorm.weight": "model.safetensors",
760
+ "model.layers.8.mlp.down_proj.biases": "model.safetensors",
761
+ "model.layers.8.mlp.down_proj.scales": "model.safetensors",
762
+ "model.layers.8.mlp.down_proj.weight": "model.safetensors",
763
+ "model.layers.8.mlp.gate_proj.biases": "model.safetensors",
764
+ "model.layers.8.mlp.gate_proj.scales": "model.safetensors",
765
+ "model.layers.8.mlp.gate_proj.weight": "model.safetensors",
766
+ "model.layers.8.mlp.up_proj.biases": "model.safetensors",
767
+ "model.layers.8.mlp.up_proj.scales": "model.safetensors",
768
+ "model.layers.8.mlp.up_proj.weight": "model.safetensors",
769
+ "model.layers.8.post_attention_layernorm.weight": "model.safetensors",
770
+ "model.layers.8.self_attn.k_proj.biases": "model.safetensors",
771
+ "model.layers.8.self_attn.k_proj.scales": "model.safetensors",
772
+ "model.layers.8.self_attn.k_proj.weight": "model.safetensors",
773
+ "model.layers.8.self_attn.key_layernorm.weight": "model.safetensors",
774
+ "model.layers.8.self_attn.o_proj.biases": "model.safetensors",
775
+ "model.layers.8.self_attn.o_proj.scales": "model.safetensors",
776
+ "model.layers.8.self_attn.o_proj.weight": "model.safetensors",
777
+ "model.layers.8.self_attn.q_proj.biases": "model.safetensors",
778
+ "model.layers.8.self_attn.q_proj.scales": "model.safetensors",
779
+ "model.layers.8.self_attn.q_proj.weight": "model.safetensors",
780
+ "model.layers.8.self_attn.query_layernorm.weight": "model.safetensors",
781
+ "model.layers.8.self_attn.v_proj.biases": "model.safetensors",
782
+ "model.layers.8.self_attn.v_proj.scales": "model.safetensors",
783
+ "model.layers.8.self_attn.v_proj.weight": "model.safetensors",
784
+ "model.layers.9.input_layernorm.weight": "model.safetensors",
785
+ "model.layers.9.mlp.down_proj.biases": "model.safetensors",
786
+ "model.layers.9.mlp.down_proj.scales": "model.safetensors",
787
+ "model.layers.9.mlp.down_proj.weight": "model.safetensors",
788
+ "model.layers.9.mlp.gate_proj.biases": "model.safetensors",
789
+ "model.layers.9.mlp.gate_proj.scales": "model.safetensors",
790
+ "model.layers.9.mlp.gate_proj.weight": "model.safetensors",
791
+ "model.layers.9.mlp.up_proj.biases": "model.safetensors",
792
+ "model.layers.9.mlp.up_proj.scales": "model.safetensors",
793
+ "model.layers.9.mlp.up_proj.weight": "model.safetensors",
794
+ "model.layers.9.post_attention_layernorm.weight": "model.safetensors",
795
+ "model.layers.9.self_attn.k_proj.biases": "model.safetensors",
796
+ "model.layers.9.self_attn.k_proj.scales": "model.safetensors",
797
+ "model.layers.9.self_attn.k_proj.weight": "model.safetensors",
798
+ "model.layers.9.self_attn.key_layernorm.weight": "model.safetensors",
799
+ "model.layers.9.self_attn.o_proj.biases": "model.safetensors",
800
+ "model.layers.9.self_attn.o_proj.scales": "model.safetensors",
801
+ "model.layers.9.self_attn.o_proj.weight": "model.safetensors",
802
+ "model.layers.9.self_attn.q_proj.biases": "model.safetensors",
803
+ "model.layers.9.self_attn.q_proj.scales": "model.safetensors",
804
+ "model.layers.9.self_attn.q_proj.weight": "model.safetensors",
805
+ "model.layers.9.self_attn.query_layernorm.weight": "model.safetensors",
806
+ "model.layers.9.self_attn.v_proj.biases": "model.safetensors",
807
+ "model.layers.9.self_attn.v_proj.scales": "model.safetensors",
808
+ "model.layers.9.self_attn.v_proj.weight": "model.safetensors",
809
+ "model.norm.weight": "model.safetensors"
810
+ }
811
+ }
modeling_hunyuan.py ADDED
The diff for this file is too large to render. See raw diff
 
special_tokens_map.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|startoftext|>",
4
+ "<|extra_0|>",
5
+ "<|extra_4|>",
6
+ "<|extra_5|>",
7
+ "<|eos|>"
8
+ ]
9
+ }
tokenization_hy.py ADDED
@@ -0,0 +1,296 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import base64
2
+ import logging
3
+ import os
4
+ import unicodedata
5
+ from typing import Collection, Dict, List, Set, Tuple, Union
6
+
7
+ import tiktoken
8
+ from transformers import PreTrainedTokenizer, AddedToken
9
+
10
+ logger = logging.getLogger(__name__)
11
+
12
+
13
+ VOCAB_FILES_NAMES = {"vocab_file": "hy.tiktoken"}
14
+
15
+ PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
16
+ # PAT_STR = r"""(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\r\n\p{L}\p{N}]?\p{L}+|\p{N}{1,3}| ?[^\s\p{L}\p{N}]+[\r\n]*|\s*[\r\n]+|\s+(?!\S)|\s+"""
17
+ ENDOFTEXT = "<|endoftext|>"
18
+ STARTOFTEXT = "<|startoftext|>"
19
+ BOSTOKEN = "<|bos|>"
20
+ EOSTOKEN = "<|eos|>"
21
+ PADTOKEN = "<|pad|>"
22
+
23
+ # as the default behavior is changed to allow special tokens in
24
+ # regular texts, the surface forms of special tokens need to be
25
+ # as different as possible to minimize the impact
26
+ EXTRAS = tuple((f"<|extra_{i}|>" for i in range(205)))
27
+ # changed to use actual index to avoid misconfiguration with vocabulary expansion
28
+
29
+
30
+ SPECIAL_START_ID = 127957
31
+
32
+ def _load_tiktoken_bpe(tiktoken_bpe_file: str) -> Dict[bytes, int]:
33
+ # with open(tiktoken_bpe_file, "rb") as f:
34
+ # contents = f.read()
35
+ dic = {}
36
+ rank = 0
37
+ for line in open(tiktoken_bpe_file, "rb"):
38
+ if line:
39
+ token, _ = line.split()
40
+ if base64.b64decode(token) in dic:
41
+ continue
42
+ dic[base64.b64decode(token)] = int(rank)
43
+ rank += 1
44
+ global SPECIAL_START_ID
45
+ SPECIAL_START_ID=rank
46
+ return dic
47
+
48
+ # print(SPECIAL_START_ID)
49
+
50
+ SPECIAL_TOKENS = tuple(
51
+ enumerate(
52
+ (
53
+ (
54
+ ENDOFTEXT,
55
+ STARTOFTEXT,
56
+ BOSTOKEN,
57
+ EOSTOKEN,
58
+ PADTOKEN,
59
+ )
60
+ + EXTRAS
61
+ ),
62
+ start=SPECIAL_START_ID,
63
+ )
64
+ )
65
+ # NOTE: Unused Token ID starts from 127962
66
+ SPECIAL_TOKENS_SET = set(t for i, t in SPECIAL_TOKENS)
67
+
68
+ class HYTokenizer(PreTrainedTokenizer):
69
+ """hunyuan tokenizer."""
70
+
71
+ vocab_files_names = VOCAB_FILES_NAMES
72
+
73
+ def __init__(
74
+ self,
75
+ vocab_file,
76
+ errors="replace",
77
+ extra_vocab_file=None,
78
+ **kwargs,
79
+ ):
80
+ super().__init__(**kwargs)
81
+
82
+ # how to handle errors in decoding UTF-8 byte sequences
83
+ # use ignore if you are in streaming inference
84
+ self.errors = errors
85
+
86
+ self.mergeable_ranks = _load_tiktoken_bpe(vocab_file) # type: Dict[bytes, int]
87
+ self.special_tokens = {
88
+ token: index
89
+ for index, token in SPECIAL_TOKENS
90
+ }
91
+
92
+ # try load extra vocab from file
93
+ if extra_vocab_file is not None:
94
+ used_ids = set(self.mergeable_ranks.values()) | set(self.special_tokens.values())
95
+ extra_mergeable_ranks = _load_tiktoken_bpe(extra_vocab_file)
96
+ for token, index in extra_mergeable_ranks.items():
97
+ if token in self.mergeable_ranks:
98
+ logger.info(f"extra token {token} exists, skipping")
99
+ continue
100
+ if index in used_ids:
101
+ logger.info(f'the index {index} for extra token {token} exists, skipping')
102
+ continue
103
+ self.mergeable_ranks[token] = index
104
+ # the index may be sparse after this, but don't worry tiktoken.Encoding will handle this
105
+
106
+ enc = tiktoken.Encoding(
107
+ "HunYuan",
108
+ pat_str=PAT_STR,
109
+ mergeable_ranks=self.mergeable_ranks,
110
+ special_tokens=self.special_tokens,
111
+ )
112
+ assert (
113
+ len(self.mergeable_ranks) + len(self.special_tokens) == enc.n_vocab
114
+ ), f"{len(self.mergeable_ranks)} + {len(self.special_tokens)} != {enc.n_vocab} in encoding"
115
+
116
+ self.decoder = {
117
+ v: k for k, v in self.mergeable_ranks.items()
118
+ } # type: dict[int, bytes|str]
119
+ self.decoder.update({v: k for k, v in self.special_tokens.items()})
120
+
121
+ self.tokenizer = enc # type: tiktoken.Encoding
122
+
123
+ self.eod_id = self.tokenizer.eot_token
124
+ self.bod_id = self.special_tokens[STARTOFTEXT]
125
+ self.bos_id = self.special_tokens[BOSTOKEN]
126
+ self.eos_id = self.special_tokens[EOSTOKEN]
127
+ self.pad_id = self.special_tokens[PADTOKEN]
128
+
129
+ def __getstate__(self):
130
+ # for pickle lovers
131
+ state = self.__dict__.copy()
132
+ del state["tokenizer"]
133
+ return state
134
+
135
+ def __setstate__(self, state):
136
+ # tokenizer is not python native; don't pass it; rebuild it
137
+ self.__dict__.update(state)
138
+ enc = tiktoken.Encoding(
139
+ "HunYuan",
140
+ pat_str=PAT_STR,
141
+ mergeable_ranks=self.mergeable_ranks,
142
+ special_tokens=self.special_tokens,
143
+ )
144
+ self.tokenizer = enc
145
+
146
+ def __len__(self) -> int:
147
+ return self.tokenizer.n_vocab
148
+
149
+ def get_vocab(self) -> Dict[bytes, int]:
150
+ return self.mergeable_ranks
151
+
152
+ def convert_tokens_to_ids(
153
+ self, tokens: Union[bytes, str, List[Union[bytes, str]]]
154
+ ) -> List[int]:
155
+ ids = []
156
+ if isinstance(tokens, (str, bytes)):
157
+ if tokens in self.special_tokens:
158
+ return self.special_tokens[tokens]
159
+ else:
160
+ return self.mergeable_ranks.get(tokens)
161
+ for token in tokens:
162
+ if token in self.special_tokens:
163
+ ids.append(self.special_tokens[token])
164
+ else:
165
+ ids.append(self.mergeable_ranks.get(token))
166
+ return ids
167
+
168
+ def _add_tokens(
169
+ self,
170
+ new_tokens: Union[List[str], List[AddedToken]],
171
+ special_tokens: bool = False,
172
+ ) -> int:
173
+ if not special_tokens and new_tokens:
174
+ raise ValueError("Adding regular tokens is not supported")
175
+ for token in new_tokens:
176
+ surface_form = token.content if isinstance(token, AddedToken) else token
177
+ if surface_form not in SPECIAL_TOKENS_SET:
178
+ raise ValueError("Adding unknown special tokens is not supported")
179
+ return 0
180
+
181
+ def save_vocabulary(self, save_directory: str, **kwargs) -> Tuple[str]:
182
+ """
183
+ Save only the vocabulary of the tokenizer (vocabulary).
184
+ Returns:
185
+ `Tuple(str)`: Paths to the files saved.
186
+ """
187
+ file_path = os.path.join(save_directory, "hunyuan.tiktoken")
188
+ with open(file_path, "w", encoding="utf8") as w:
189
+ for k, v in self.mergeable_ranks.items():
190
+ line = base64.b64encode(k).decode("utf8") + " " + str(v) + "\n"
191
+ w.write(line)
192
+ return (file_path,)
193
+
194
+ def tokenize(
195
+ self,
196
+ text: str,
197
+ allowed_special: Union[Set, str] = "all",
198
+ disallowed_special: Union[Collection, str] = (),
199
+ **kwargs,
200
+ ) -> List[Union[bytes, str]]:
201
+ """
202
+ Converts a string in a sequence of tokens.
203
+ Args:
204
+ text (`str`):
205
+ The sequence to be encoded.
206
+ allowed_special (`Literal["all"]` or `set`):
207
+ The surface forms of the tokens to be encoded as special tokens in regular texts.
208
+ Default to "all".
209
+ disallowed_special (`Literal["all"]` or `Collection`):
210
+ The surface forms of the tokens that should not be in regular texts and trigger errors.
211
+ Default to an empty tuple.
212
+ kwargs (additional keyword arguments, *optional*):
213
+ Will be passed to the underlying model specific encode method.
214
+ Returns:
215
+ `List[bytes|str]`: The list of tokens.
216
+ """
217
+ tokens = []
218
+ text = unicodedata.normalize("NFC", text)
219
+
220
+ # this implementation takes a detour: text -> token id -> token surface forms
221
+ for t in self.tokenizer.encode(
222
+ text, allowed_special=allowed_special, disallowed_special=disallowed_special
223
+ ):
224
+ tokens.append(self.decoder[t])
225
+ return tokens
226
+
227
+ def convert_tokens_to_string(self, tokens: List[Union[bytes, str]]) -> str:
228
+ """
229
+ Converts a sequence of tokens in a single string.
230
+ """
231
+ text = ""
232
+ temp = b""
233
+ for t in tokens:
234
+ if isinstance(t, str):
235
+ if temp:
236
+ text += temp.decode("utf-8", errors=self.errors)
237
+ temp = b""
238
+ text += t
239
+ elif isinstance(t, bytes):
240
+ temp += t
241
+ else:
242
+ raise TypeError("token should only be of type types or str")
243
+ if temp:
244
+ text += temp.decode("utf-8", errors=self.errors)
245
+ return text
246
+
247
+ @property
248
+ def vocab_size(self):
249
+ return self.tokenizer.n_vocab
250
+
251
+ def _convert_id_to_token(self, index: int) -> Union[bytes, str]:
252
+ """Converts an id to a token, special tokens included"""
253
+ if index in self.decoder:
254
+ return self.decoder[index]
255
+ raise ValueError("unknown ids")
256
+
257
+ def _convert_token_to_id(self, token: Union[bytes, str]) -> int:
258
+ """Converts a token to an id using the vocab, special tokens included"""
259
+ if token in self.special_tokens:
260
+ return self.special_tokens[token]
261
+ if token in self.mergeable_ranks:
262
+ return self.mergeable_ranks[token]
263
+ raise ValueError("unknown token")
264
+
265
+ def _tokenize(self, text: str, **kwargs):
266
+ """
267
+ Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
268
+ vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
269
+ Do NOT take care of added tokens.
270
+ """
271
+ raise NotImplementedError
272
+
273
+ def _decode(
274
+ self,
275
+ token_ids: Union[int, List[int]],
276
+ skip_special_tokens: bool = False,
277
+ errors: str = None,
278
+ **kwargs,
279
+ ) -> str:
280
+ if isinstance(token_ids, int):
281
+ token_ids = [token_ids]
282
+ if skip_special_tokens:
283
+ token_ids = [i for i in token_ids if i < self.eod_id]
284
+ return self.tokenizer.decode(token_ids, errors=errors or self.errors)
285
+
286
+ # tests
287
+ if __name__ == "__main__":
288
+ tokenizer = HYTokenizer.from_pretrained('/hy')
289
+ text = '你好，世界'
290
+ tokens = tokenizer.tokenize(text)
291
+ print(tokens)
292
+ ids = tokenizer.convert_tokens_to_ids(tokens)
293
+ print(ids)
294
+ text2 = tokenizer.convert_tokens_to_string(tokens)
295
+ print(text2)
296
+ ids2 = tokenizer.convert_tokens_to_ids(tokens)
tokenizer_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {},
3
+ "additional_special_tokens": [
4
+ "<|startoftext|>",
5
+ "<|extra_0|>",
6
+ "<|extra_4|>",
7
+ "<|extra_5|>",
8
+ "<|eos|>"
9
+ ],
10
+ "auto_map": {
11
+ "AutoTokenizer": [
12
+ "tokenization_hy.HYTokenizer",
13
+ null
14
+ ]
15
+ },
16
+ "chat_template": "{% set context = {'has_head': true} %}{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = message['content'] %}{% if loop.index0 == 0 %}{% if content == '' %}{% set _ = context.update({'has_head': false}) %}{% else %}{% set content = '<|startoftext|>' + content + '<|extra_4|>' %}{% endif %}{% endif %}{% if message['role'] == 'user' %}{% if loop.index0 == 1 and not context.has_head %}{% set content = '<|startoftext|>' + content %}{% endif %}{% if loop.index0 == 1 and context.has_head %}{% set content = content + '<|extra_0|>' %}{% else %}{% set content = '<|startoftext|>' + content + '<|extra_0|>' %}{% endif %}{% elif message['role'] == 'assistant' %}{% set content = content + '<|eos|>' %}{% endif %}{{ content }}{% endfor %}",
17
+ "clean_up_tokenization_spaces": false,
18
+ "extra_special_tokens": {},
19
+ "model_max_length": 1048576,
20
+ "tokenizer_class": "HYTokenizer"
21
+ }