nightmedia commited on
Commit
0e54c70
·
verified ·
1 Parent(s): 5bd4711

Add files using upload-large-folder tool

Browse files
README.md CHANGED
@@ -1,3 +1,54 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: other
3
+ license_name: mnpl
4
+ license_link: https://mistral.ai/licences/MNPL-0.1.md
5
+ language:
6
+ - en
7
+ base_model: DavidAU/Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B
8
+ pipeline_tag: text-generation
9
+ tags:
10
+ - merge
11
+ - programming
12
+ - code generation
13
+ - Codestral
14
+ - code
15
+ - moe
16
+ - coding
17
+ - coder
18
+ - chat
19
+ - mistral
20
+ - mixtral
21
+ - mixture of experts
22
+ - mistral moe
23
+ - 2X22B
24
+ - mlx
25
+ library_name: mlx
26
+ ---
27
+
28
+ # Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B-q4-mlx
29
+
30
+ This model [Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B-q4-mlx](https://huggingface.co/Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B-q4-mlx) was
31
+ converted to MLX format from [DavidAU/Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B](https://huggingface.co/DavidAU/Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B)
32
+ using mlx-lm version **0.25.3**.
33
+
34
+ ## Use with mlx
35
+
36
+ ```bash
37
+ pip install mlx-lm
38
+ ```
39
+
40
+ ```python
41
+ from mlx_lm import load, generate
42
+
43
+ model, tokenizer = load("Mistral-2x22B-MOE-Power-Codestral-Ultimate-39B-q4-mlx")
44
+
45
+ prompt = "hello"
46
+
47
+ if tokenizer.chat_template is not None:
48
+ messages = [{"role": "user", "content": prompt}]
49
+ prompt = tokenizer.apply_chat_template(
50
+ messages, add_generation_prompt=True
51
+ )
52
+
53
+ response = generate(model, tokenizer, prompt=prompt, verbose=True)
54
+ ```
chat_template.jinja ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '
2
+ ' + message['content'] + '<|im_end|>' + '
3
+ '}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
4
+ ' }}{% endif %})
config.json ADDED
@@ -0,0 +1,1389 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MixtralForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "head_dim": null,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 6144,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 16384,
13
+ "max_position_embeddings": 32768,
14
+ "model_type": "mixtral",
15
+ "num_attention_heads": 48,
16
+ "num_experts_per_tok": 2,
17
+ "num_hidden_layers": 56,
18
+ "num_key_value_heads": 8,
19
+ "num_local_experts": 2,
20
+ "output_router_logits": false,
21
+ "quantization": {
22
+ "group_size": 64,
23
+ "bits": 4,
24
+ "model.embed_tokens": true,
25
+ "model.layers.0.self_attn.q_proj": true,
26
+ "model.layers.0.self_attn.k_proj": true,
27
+ "model.layers.0.self_attn.v_proj": true,
28
+ "model.layers.0.self_attn.o_proj": true,
29
+ "model.layers.0.self_attn.rope": false,
30
+ "model.layers.0.block_sparse_moe.gate": true,
31
+ "model.layers.0.block_sparse_moe.switch_mlp.gate_proj": true,
32
+ "model.layers.0.block_sparse_moe.switch_mlp.up_proj": true,
33
+ "model.layers.0.block_sparse_moe.switch_mlp.down_proj": true,
34
+ "model.layers.0.block_sparse_moe.switch_mlp.activation": false,
35
+ "model.layers.0.input_layernorm": false,
36
+ "model.layers.0.post_attention_layernorm": false,
37
+ "model.layers.1.self_attn.q_proj": true,
38
+ "model.layers.1.self_attn.k_proj": true,
39
+ "model.layers.1.self_attn.v_proj": true,
40
+ "model.layers.1.self_attn.o_proj": true,
41
+ "model.layers.1.self_attn.rope": false,
42
+ "model.layers.1.block_sparse_moe.gate": true,
43
+ "model.layers.1.block_sparse_moe.switch_mlp.gate_proj": true,
44
+ "model.layers.1.block_sparse_moe.switch_mlp.up_proj": true,
45
+ "model.layers.1.block_sparse_moe.switch_mlp.down_proj": true,
46
+ "model.layers.1.block_sparse_moe.switch_mlp.activation": false,
47
+ "model.layers.1.input_layernorm": false,
48
+ "model.layers.1.post_attention_layernorm": false,
49
+ "model.layers.2.self_attn.q_proj": true,
50
+ "model.layers.2.self_attn.k_proj": true,
51
+ "model.layers.2.self_attn.v_proj": true,
52
+ "model.layers.2.self_attn.o_proj": true,
53
+ "model.layers.2.self_attn.rope": false,
54
+ "model.layers.2.block_sparse_moe.gate": true,
55
+ "model.layers.2.block_sparse_moe.switch_mlp.gate_proj": true,
56
+ "model.layers.2.block_sparse_moe.switch_mlp.up_proj": true,
57
+ "model.layers.2.block_sparse_moe.switch_mlp.down_proj": true,
58
+ "model.layers.2.block_sparse_moe.switch_mlp.activation": false,
59
+ "model.layers.2.input_layernorm": false,
60
+ "model.layers.2.post_attention_layernorm": false,
61
+ "model.layers.3.self_attn.q_proj": true,
62
+ "model.layers.3.self_attn.k_proj": true,
63
+ "model.layers.3.self_attn.v_proj": true,
64
+ "model.layers.3.self_attn.o_proj": true,
65
+ "model.layers.3.self_attn.rope": false,
66
+ "model.layers.3.block_sparse_moe.gate": true,
67
+ "model.layers.3.block_sparse_moe.switch_mlp.gate_proj": true,
68
+ "model.layers.3.block_sparse_moe.switch_mlp.up_proj": true,
69
+ "model.layers.3.block_sparse_moe.switch_mlp.down_proj": true,
70
+ "model.layers.3.block_sparse_moe.switch_mlp.activation": false,
71
+ "model.layers.3.input_layernorm": false,
72
+ "model.layers.3.post_attention_layernorm": false,
73
+ "model.layers.4.self_attn.q_proj": true,
74
+ "model.layers.4.self_attn.k_proj": true,
75
+ "model.layers.4.self_attn.v_proj": true,
76
+ "model.layers.4.self_attn.o_proj": true,
77
+ "model.layers.4.self_attn.rope": false,
78
+ "model.layers.4.block_sparse_moe.gate": true,
79
+ "model.layers.4.block_sparse_moe.switch_mlp.gate_proj": true,
80
+ "model.layers.4.block_sparse_moe.switch_mlp.up_proj": true,
81
+ "model.layers.4.block_sparse_moe.switch_mlp.down_proj": true,
82
+ "model.layers.4.block_sparse_moe.switch_mlp.activation": false,
83
+ "model.layers.4.input_layernorm": false,
84
+ "model.layers.4.post_attention_layernorm": false,
85
+ "model.layers.5.self_attn.q_proj": true,
86
+ "model.layers.5.self_attn.k_proj": true,
87
+ "model.layers.5.self_attn.v_proj": true,
88
+ "model.layers.5.self_attn.o_proj": true,
89
+ "model.layers.5.self_attn.rope": false,
90
+ "model.layers.5.block_sparse_moe.gate": true,
91
+ "model.layers.5.block_sparse_moe.switch_mlp.gate_proj": true,
92
+ "model.layers.5.block_sparse_moe.switch_mlp.up_proj": true,
93
+ "model.layers.5.block_sparse_moe.switch_mlp.down_proj": true,
94
+ "model.layers.5.block_sparse_moe.switch_mlp.activation": false,
95
+ "model.layers.5.input_layernorm": false,
96
+ "model.layers.5.post_attention_layernorm": false,
97
+ "model.layers.6.self_attn.q_proj": true,
98
+ "model.layers.6.self_attn.k_proj": true,
99
+ "model.layers.6.self_attn.v_proj": true,
100
+ "model.layers.6.self_attn.o_proj": true,
101
+ "model.layers.6.self_attn.rope": false,
102
+ "model.layers.6.block_sparse_moe.gate": true,
103
+ "model.layers.6.block_sparse_moe.switch_mlp.gate_proj": true,
104
+ "model.layers.6.block_sparse_moe.switch_mlp.up_proj": true,
105
+ "model.layers.6.block_sparse_moe.switch_mlp.down_proj": true,
106
+ "model.layers.6.block_sparse_moe.switch_mlp.activation": false,
107
+ "model.layers.6.input_layernorm": false,
108
+ "model.layers.6.post_attention_layernorm": false,
109
+ "model.layers.7.self_attn.q_proj": true,
110
+ "model.layers.7.self_attn.k_proj": true,
111
+ "model.layers.7.self_attn.v_proj": true,
112
+ "model.layers.7.self_attn.o_proj": true,
113
+ "model.layers.7.self_attn.rope": false,
114
+ "model.layers.7.block_sparse_moe.gate": true,
115
+ "model.layers.7.block_sparse_moe.switch_mlp.gate_proj": true,
116
+ "model.layers.7.block_sparse_moe.switch_mlp.up_proj": true,
117
+ "model.layers.7.block_sparse_moe.switch_mlp.down_proj": true,
118
+ "model.layers.7.block_sparse_moe.switch_mlp.activation": false,
119
+ "model.layers.7.input_layernorm": false,
120
+ "model.layers.7.post_attention_layernorm": false,
121
+ "model.layers.8.self_attn.q_proj": true,
122
+ "model.layers.8.self_attn.k_proj": true,
123
+ "model.layers.8.self_attn.v_proj": true,
124
+ "model.layers.8.self_attn.o_proj": true,
125
+ "model.layers.8.self_attn.rope": false,
126
+ "model.layers.8.block_sparse_moe.gate": true,
127
+ "model.layers.8.block_sparse_moe.switch_mlp.gate_proj": true,
128
+ "model.layers.8.block_sparse_moe.switch_mlp.up_proj": true,
129
+ "model.layers.8.block_sparse_moe.switch_mlp.down_proj": true,
130
+ "model.layers.8.block_sparse_moe.switch_mlp.activation": false,
131
+ "model.layers.8.input_layernorm": false,
132
+ "model.layers.8.post_attention_layernorm": false,
133
+ "model.layers.9.self_attn.q_proj": true,
134
+ "model.layers.9.self_attn.k_proj": true,
135
+ "model.layers.9.self_attn.v_proj": true,
136
+ "model.layers.9.self_attn.o_proj": true,
137
+ "model.layers.9.self_attn.rope": false,
138
+ "model.layers.9.block_sparse_moe.gate": true,
139
+ "model.layers.9.block_sparse_moe.switch_mlp.gate_proj": true,
140
+ "model.layers.9.block_sparse_moe.switch_mlp.up_proj": true,
141
+ "model.layers.9.block_sparse_moe.switch_mlp.down_proj": true,
142
+ "model.layers.9.block_sparse_moe.switch_mlp.activation": false,
143
+ "model.layers.9.input_layernorm": false,
144
+ "model.layers.9.post_attention_layernorm": false,
145
+ "model.layers.10.self_attn.q_proj": true,
146
+ "model.layers.10.self_attn.k_proj": true,
147
+ "model.layers.10.self_attn.v_proj": true,
148
+ "model.layers.10.self_attn.o_proj": true,
149
+ "model.layers.10.self_attn.rope": false,
150
+ "model.layers.10.block_sparse_moe.gate": true,
151
+ "model.layers.10.block_sparse_moe.switch_mlp.gate_proj": true,
152
+ "model.layers.10.block_sparse_moe.switch_mlp.up_proj": true,
153
+ "model.layers.10.block_sparse_moe.switch_mlp.down_proj": true,
154
+ "model.layers.10.block_sparse_moe.switch_mlp.activation": false,
155
+ "model.layers.10.input_layernorm": false,
156
+ "model.layers.10.post_attention_layernorm": false,
157
+ "model.layers.11.self_attn.q_proj": true,
158
+ "model.layers.11.self_attn.k_proj": true,
159
+ "model.layers.11.self_attn.v_proj": true,
160
+ "model.layers.11.self_attn.o_proj": true,
161
+ "model.layers.11.self_attn.rope": false,
162
+ "model.layers.11.block_sparse_moe.gate": true,
163
+ "model.layers.11.block_sparse_moe.switch_mlp.gate_proj": true,
164
+ "model.layers.11.block_sparse_moe.switch_mlp.up_proj": true,
165
+ "model.layers.11.block_sparse_moe.switch_mlp.down_proj": true,
166
+ "model.layers.11.block_sparse_moe.switch_mlp.activation": false,
167
+ "model.layers.11.input_layernorm": false,
168
+ "model.layers.11.post_attention_layernorm": false,
169
+ "model.layers.12.self_attn.q_proj": true,
170
+ "model.layers.12.self_attn.k_proj": true,
171
+ "model.layers.12.self_attn.v_proj": true,
172
+ "model.layers.12.self_attn.o_proj": true,
173
+ "model.layers.12.self_attn.rope": false,
174
+ "model.layers.12.block_sparse_moe.gate": true,
175
+ "model.layers.12.block_sparse_moe.switch_mlp.gate_proj": true,
176
+ "model.layers.12.block_sparse_moe.switch_mlp.up_proj": true,
177
+ "model.layers.12.block_sparse_moe.switch_mlp.down_proj": true,
178
+ "model.layers.12.block_sparse_moe.switch_mlp.activation": false,
179
+ "model.layers.12.input_layernorm": false,
180
+ "model.layers.12.post_attention_layernorm": false,
181
+ "model.layers.13.self_attn.q_proj": true,
182
+ "model.layers.13.self_attn.k_proj": true,
183
+ "model.layers.13.self_attn.v_proj": true,
184
+ "model.layers.13.self_attn.o_proj": true,
185
+ "model.layers.13.self_attn.rope": false,
186
+ "model.layers.13.block_sparse_moe.gate": true,
187
+ "model.layers.13.block_sparse_moe.switch_mlp.gate_proj": true,
188
+ "model.layers.13.block_sparse_moe.switch_mlp.up_proj": true,
189
+ "model.layers.13.block_sparse_moe.switch_mlp.down_proj": true,
190
+ "model.layers.13.block_sparse_moe.switch_mlp.activation": false,
191
+ "model.layers.13.input_layernorm": false,
192
+ "model.layers.13.post_attention_layernorm": false,
193
+ "model.layers.14.self_attn.q_proj": true,
194
+ "model.layers.14.self_attn.k_proj": true,
195
+ "model.layers.14.self_attn.v_proj": true,
196
+ "model.layers.14.self_attn.o_proj": true,
197
+ "model.layers.14.self_attn.rope": false,
198
+ "model.layers.14.block_sparse_moe.gate": true,
199
+ "model.layers.14.block_sparse_moe.switch_mlp.gate_proj": true,
200
+ "model.layers.14.block_sparse_moe.switch_mlp.up_proj": true,
201
+ "model.layers.14.block_sparse_moe.switch_mlp.down_proj": true,
202
+ "model.layers.14.block_sparse_moe.switch_mlp.activation": false,
203
+ "model.layers.14.input_layernorm": false,
204
+ "model.layers.14.post_attention_layernorm": false,
205
+ "model.layers.15.self_attn.q_proj": true,
206
+ "model.layers.15.self_attn.k_proj": true,
207
+ "model.layers.15.self_attn.v_proj": true,
208
+ "model.layers.15.self_attn.o_proj": true,
209
+ "model.layers.15.self_attn.rope": false,
210
+ "model.layers.15.block_sparse_moe.gate": true,
211
+ "model.layers.15.block_sparse_moe.switch_mlp.gate_proj": true,
212
+ "model.layers.15.block_sparse_moe.switch_mlp.up_proj": true,
213
+ "model.layers.15.block_sparse_moe.switch_mlp.down_proj": true,
214
+ "model.layers.15.block_sparse_moe.switch_mlp.activation": false,
215
+ "model.layers.15.input_layernorm": false,
216
+ "model.layers.15.post_attention_layernorm": false,
217
+ "model.layers.16.self_attn.q_proj": true,
218
+ "model.layers.16.self_attn.k_proj": true,
219
+ "model.layers.16.self_attn.v_proj": true,
220
+ "model.layers.16.self_attn.o_proj": true,
221
+ "model.layers.16.self_attn.rope": false,
222
+ "model.layers.16.block_sparse_moe.gate": true,
223
+ "model.layers.16.block_sparse_moe.switch_mlp.gate_proj": true,
224
+ "model.layers.16.block_sparse_moe.switch_mlp.up_proj": true,
225
+ "model.layers.16.block_sparse_moe.switch_mlp.down_proj": true,
226
+ "model.layers.16.block_sparse_moe.switch_mlp.activation": false,
227
+ "model.layers.16.input_layernorm": false,
228
+ "model.layers.16.post_attention_layernorm": false,
229
+ "model.layers.17.self_attn.q_proj": true,
230
+ "model.layers.17.self_attn.k_proj": true,
231
+ "model.layers.17.self_attn.v_proj": true,
232
+ "model.layers.17.self_attn.o_proj": true,
233
+ "model.layers.17.self_attn.rope": false,
234
+ "model.layers.17.block_sparse_moe.gate": true,
235
+ "model.layers.17.block_sparse_moe.switch_mlp.gate_proj": true,
236
+ "model.layers.17.block_sparse_moe.switch_mlp.up_proj": true,
237
+ "model.layers.17.block_sparse_moe.switch_mlp.down_proj": true,
238
+ "model.layers.17.block_sparse_moe.switch_mlp.activation": false,
239
+ "model.layers.17.input_layernorm": false,
240
+ "model.layers.17.post_attention_layernorm": false,
241
+ "model.layers.18.self_attn.q_proj": true,
242
+ "model.layers.18.self_attn.k_proj": true,
243
+ "model.layers.18.self_attn.v_proj": true,
244
+ "model.layers.18.self_attn.o_proj": true,
245
+ "model.layers.18.self_attn.rope": false,
246
+ "model.layers.18.block_sparse_moe.gate": true,
247
+ "model.layers.18.block_sparse_moe.switch_mlp.gate_proj": true,
248
+ "model.layers.18.block_sparse_moe.switch_mlp.up_proj": true,
249
+ "model.layers.18.block_sparse_moe.switch_mlp.down_proj": true,
250
+ "model.layers.18.block_sparse_moe.switch_mlp.activation": false,
251
+ "model.layers.18.input_layernorm": false,
252
+ "model.layers.18.post_attention_layernorm": false,
253
+ "model.layers.19.self_attn.q_proj": true,
254
+ "model.layers.19.self_attn.k_proj": true,
255
+ "model.layers.19.self_attn.v_proj": true,
256
+ "model.layers.19.self_attn.o_proj": true,
257
+ "model.layers.19.self_attn.rope": false,
258
+ "model.layers.19.block_sparse_moe.gate": true,
259
+ "model.layers.19.block_sparse_moe.switch_mlp.gate_proj": true,
260
+ "model.layers.19.block_sparse_moe.switch_mlp.up_proj": true,
261
+ "model.layers.19.block_sparse_moe.switch_mlp.down_proj": true,
262
+ "model.layers.19.block_sparse_moe.switch_mlp.activation": false,
263
+ "model.layers.19.input_layernorm": false,
264
+ "model.layers.19.post_attention_layernorm": false,
265
+ "model.layers.20.self_attn.q_proj": true,
266
+ "model.layers.20.self_attn.k_proj": true,
267
+ "model.layers.20.self_attn.v_proj": true,
268
+ "model.layers.20.self_attn.o_proj": true,
269
+ "model.layers.20.self_attn.rope": false,
270
+ "model.layers.20.block_sparse_moe.gate": true,
271
+ "model.layers.20.block_sparse_moe.switch_mlp.gate_proj": true,
272
+ "model.layers.20.block_sparse_moe.switch_mlp.up_proj": true,
273
+ "model.layers.20.block_sparse_moe.switch_mlp.down_proj": true,
274
+ "model.layers.20.block_sparse_moe.switch_mlp.activation": false,
275
+ "model.layers.20.input_layernorm": false,
276
+ "model.layers.20.post_attention_layernorm": false,
277
+ "model.layers.21.self_attn.q_proj": true,
278
+ "model.layers.21.self_attn.k_proj": true,
279
+ "model.layers.21.self_attn.v_proj": true,
280
+ "model.layers.21.self_attn.o_proj": true,
281
+ "model.layers.21.self_attn.rope": false,
282
+ "model.layers.21.block_sparse_moe.gate": true,
283
+ "model.layers.21.block_sparse_moe.switch_mlp.gate_proj": true,
284
+ "model.layers.21.block_sparse_moe.switch_mlp.up_proj": true,
285
+ "model.layers.21.block_sparse_moe.switch_mlp.down_proj": true,
286
+ "model.layers.21.block_sparse_moe.switch_mlp.activation": false,
287
+ "model.layers.21.input_layernorm": false,
288
+ "model.layers.21.post_attention_layernorm": false,
289
+ "model.layers.22.self_attn.q_proj": true,
290
+ "model.layers.22.self_attn.k_proj": true,
291
+ "model.layers.22.self_attn.v_proj": true,
292
+ "model.layers.22.self_attn.o_proj": true,
293
+ "model.layers.22.self_attn.rope": false,
294
+ "model.layers.22.block_sparse_moe.gate": true,
295
+ "model.layers.22.block_sparse_moe.switch_mlp.gate_proj": true,
296
+ "model.layers.22.block_sparse_moe.switch_mlp.up_proj": true,
297
+ "model.layers.22.block_sparse_moe.switch_mlp.down_proj": true,
298
+ "model.layers.22.block_sparse_moe.switch_mlp.activation": false,
299
+ "model.layers.22.input_layernorm": false,
300
+ "model.layers.22.post_attention_layernorm": false,
301
+ "model.layers.23.self_attn.q_proj": true,
302
+ "model.layers.23.self_attn.k_proj": true,
303
+ "model.layers.23.self_attn.v_proj": true,
304
+ "model.layers.23.self_attn.o_proj": true,
305
+ "model.layers.23.self_attn.rope": false,
306
+ "model.layers.23.block_sparse_moe.gate": true,
307
+ "model.layers.23.block_sparse_moe.switch_mlp.gate_proj": true,
308
+ "model.layers.23.block_sparse_moe.switch_mlp.up_proj": true,
309
+ "model.layers.23.block_sparse_moe.switch_mlp.down_proj": true,
310
+ "model.layers.23.block_sparse_moe.switch_mlp.activation": false,
311
+ "model.layers.23.input_layernorm": false,
312
+ "model.layers.23.post_attention_layernorm": false,
313
+ "model.layers.24.self_attn.q_proj": true,
314
+ "model.layers.24.self_attn.k_proj": true,
315
+ "model.layers.24.self_attn.v_proj": true,
316
+ "model.layers.24.self_attn.o_proj": true,
317
+ "model.layers.24.self_attn.rope": false,
318
+ "model.layers.24.block_sparse_moe.gate": true,
319
+ "model.layers.24.block_sparse_moe.switch_mlp.gate_proj": true,
320
+ "model.layers.24.block_sparse_moe.switch_mlp.up_proj": true,
321
+ "model.layers.24.block_sparse_moe.switch_mlp.down_proj": true,
322
+ "model.layers.24.block_sparse_moe.switch_mlp.activation": false,
323
+ "model.layers.24.input_layernorm": false,
324
+ "model.layers.24.post_attention_layernorm": false,
325
+ "model.layers.25.self_attn.q_proj": true,
326
+ "model.layers.25.self_attn.k_proj": true,
327
+ "model.layers.25.self_attn.v_proj": true,
328
+ "model.layers.25.self_attn.o_proj": true,
329
+ "model.layers.25.self_attn.rope": false,
330
+ "model.layers.25.block_sparse_moe.gate": true,
331
+ "model.layers.25.block_sparse_moe.switch_mlp.gate_proj": true,
332
+ "model.layers.25.block_sparse_moe.switch_mlp.up_proj": true,
333
+ "model.layers.25.block_sparse_moe.switch_mlp.down_proj": true,
334
+ "model.layers.25.block_sparse_moe.switch_mlp.activation": false,
335
+ "model.layers.25.input_layernorm": false,
336
+ "model.layers.25.post_attention_layernorm": false,
337
+ "model.layers.26.self_attn.q_proj": true,
338
+ "model.layers.26.self_attn.k_proj": true,
339
+ "model.layers.26.self_attn.v_proj": true,
340
+ "model.layers.26.self_attn.o_proj": true,
341
+ "model.layers.26.self_attn.rope": false,
342
+ "model.layers.26.block_sparse_moe.gate": true,
343
+ "model.layers.26.block_sparse_moe.switch_mlp.gate_proj": true,
344
+ "model.layers.26.block_sparse_moe.switch_mlp.up_proj": true,
345
+ "model.layers.26.block_sparse_moe.switch_mlp.down_proj": true,
346
+ "model.layers.26.block_sparse_moe.switch_mlp.activation": false,
347
+ "model.layers.26.input_layernorm": false,
348
+ "model.layers.26.post_attention_layernorm": false,
349
+ "model.layers.27.self_attn.q_proj": true,
350
+ "model.layers.27.self_attn.k_proj": true,
351
+ "model.layers.27.self_attn.v_proj": true,
352
+ "model.layers.27.self_attn.o_proj": true,
353
+ "model.layers.27.self_attn.rope": false,
354
+ "model.layers.27.block_sparse_moe.gate": true,
355
+ "model.layers.27.block_sparse_moe.switch_mlp.gate_proj": true,
356
+ "model.layers.27.block_sparse_moe.switch_mlp.up_proj": true,
357
+ "model.layers.27.block_sparse_moe.switch_mlp.down_proj": true,
358
+ "model.layers.27.block_sparse_moe.switch_mlp.activation": false,
359
+ "model.layers.27.input_layernorm": false,
360
+ "model.layers.27.post_attention_layernorm": false,
361
+ "model.layers.28.self_attn.q_proj": true,
362
+ "model.layers.28.self_attn.k_proj": true,
363
+ "model.layers.28.self_attn.v_proj": true,
364
+ "model.layers.28.self_attn.o_proj": true,
365
+ "model.layers.28.self_attn.rope": false,
366
+ "model.layers.28.block_sparse_moe.gate": true,
367
+ "model.layers.28.block_sparse_moe.switch_mlp.gate_proj": true,
368
+ "model.layers.28.block_sparse_moe.switch_mlp.up_proj": true,
369
+ "model.layers.28.block_sparse_moe.switch_mlp.down_proj": true,
370
+ "model.layers.28.block_sparse_moe.switch_mlp.activation": false,
371
+ "model.layers.28.input_layernorm": false,
372
+ "model.layers.28.post_attention_layernorm": false,
373
+ "model.layers.29.self_attn.q_proj": true,
374
+ "model.layers.29.self_attn.k_proj": true,
375
+ "model.layers.29.self_attn.v_proj": true,
376
+ "model.layers.29.self_attn.o_proj": true,
377
+ "model.layers.29.self_attn.rope": false,
378
+ "model.layers.29.block_sparse_moe.gate": true,
379
+ "model.layers.29.block_sparse_moe.switch_mlp.gate_proj": true,
380
+ "model.layers.29.block_sparse_moe.switch_mlp.up_proj": true,
381
+ "model.layers.29.block_sparse_moe.switch_mlp.down_proj": true,
382
+ "model.layers.29.block_sparse_moe.switch_mlp.activation": false,
383
+ "model.layers.29.input_layernorm": false,
384
+ "model.layers.29.post_attention_layernorm": false,
385
+ "model.layers.30.self_attn.q_proj": true,
386
+ "model.layers.30.self_attn.k_proj": true,
387
+ "model.layers.30.self_attn.v_proj": true,
388
+ "model.layers.30.self_attn.o_proj": true,
389
+ "model.layers.30.self_attn.rope": false,
390
+ "model.layers.30.block_sparse_moe.gate": true,
391
+ "model.layers.30.block_sparse_moe.switch_mlp.gate_proj": true,
392
+ "model.layers.30.block_sparse_moe.switch_mlp.up_proj": true,
393
+ "model.layers.30.block_sparse_moe.switch_mlp.down_proj": true,
394
+ "model.layers.30.block_sparse_moe.switch_mlp.activation": false,
395
+ "model.layers.30.input_layernorm": false,
396
+ "model.layers.30.post_attention_layernorm": false,
397
+ "model.layers.31.self_attn.q_proj": true,
398
+ "model.layers.31.self_attn.k_proj": true,
399
+ "model.layers.31.self_attn.v_proj": true,
400
+ "model.layers.31.self_attn.o_proj": true,
401
+ "model.layers.31.self_attn.rope": false,
402
+ "model.layers.31.block_sparse_moe.gate": true,
403
+ "model.layers.31.block_sparse_moe.switch_mlp.gate_proj": true,
404
+ "model.layers.31.block_sparse_moe.switch_mlp.up_proj": true,
405
+ "model.layers.31.block_sparse_moe.switch_mlp.down_proj": true,
406
+ "model.layers.31.block_sparse_moe.switch_mlp.activation": false,
407
+ "model.layers.31.input_layernorm": false,
408
+ "model.layers.31.post_attention_layernorm": false,
409
+ "model.layers.32.self_attn.q_proj": true,
410
+ "model.layers.32.self_attn.k_proj": true,
411
+ "model.layers.32.self_attn.v_proj": true,
412
+ "model.layers.32.self_attn.o_proj": true,
413
+ "model.layers.32.self_attn.rope": false,
414
+ "model.layers.32.block_sparse_moe.gate": true,
415
+ "model.layers.32.block_sparse_moe.switch_mlp.gate_proj": true,
416
+ "model.layers.32.block_sparse_moe.switch_mlp.up_proj": true,
417
+ "model.layers.32.block_sparse_moe.switch_mlp.down_proj": true,
418
+ "model.layers.32.block_sparse_moe.switch_mlp.activation": false,
419
+ "model.layers.32.input_layernorm": false,
420
+ "model.layers.32.post_attention_layernorm": false,
421
+ "model.layers.33.self_attn.q_proj": true,
422
+ "model.layers.33.self_attn.k_proj": true,
423
+ "model.layers.33.self_attn.v_proj": true,
424
+ "model.layers.33.self_attn.o_proj": true,
425
+ "model.layers.33.self_attn.rope": false,
426
+ "model.layers.33.block_sparse_moe.gate": true,
427
+ "model.layers.33.block_sparse_moe.switch_mlp.gate_proj": true,
428
+ "model.layers.33.block_sparse_moe.switch_mlp.up_proj": true,
429
+ "model.layers.33.block_sparse_moe.switch_mlp.down_proj": true,
430
+ "model.layers.33.block_sparse_moe.switch_mlp.activation": false,
431
+ "model.layers.33.input_layernorm": false,
432
+ "model.layers.33.post_attention_layernorm": false,
433
+ "model.layers.34.self_attn.q_proj": true,
434
+ "model.layers.34.self_attn.k_proj": true,
435
+ "model.layers.34.self_attn.v_proj": true,
436
+ "model.layers.34.self_attn.o_proj": true,
437
+ "model.layers.34.self_attn.rope": false,
438
+ "model.layers.34.block_sparse_moe.gate": true,
439
+ "model.layers.34.block_sparse_moe.switch_mlp.gate_proj": true,
440
+ "model.layers.34.block_sparse_moe.switch_mlp.up_proj": true,
441
+ "model.layers.34.block_sparse_moe.switch_mlp.down_proj": true,
442
+ "model.layers.34.block_sparse_moe.switch_mlp.activation": false,
443
+ "model.layers.34.input_layernorm": false,
444
+ "model.layers.34.post_attention_layernorm": false,
445
+ "model.layers.35.self_attn.q_proj": true,
446
+ "model.layers.35.self_attn.k_proj": true,
447
+ "model.layers.35.self_attn.v_proj": true,
448
+ "model.layers.35.self_attn.o_proj": true,
449
+ "model.layers.35.self_attn.rope": false,
450
+ "model.layers.35.block_sparse_moe.gate": true,
451
+ "model.layers.35.block_sparse_moe.switch_mlp.gate_proj": true,
452
+ "model.layers.35.block_sparse_moe.switch_mlp.up_proj": true,
453
+ "model.layers.35.block_sparse_moe.switch_mlp.down_proj": true,
454
+ "model.layers.35.block_sparse_moe.switch_mlp.activation": false,
455
+ "model.layers.35.input_layernorm": false,
456
+ "model.layers.35.post_attention_layernorm": false,
457
+ "model.layers.36.self_attn.q_proj": true,
458
+ "model.layers.36.self_attn.k_proj": true,
459
+ "model.layers.36.self_attn.v_proj": true,
460
+ "model.layers.36.self_attn.o_proj": true,
461
+ "model.layers.36.self_attn.rope": false,
462
+ "model.layers.36.block_sparse_moe.gate": true,
463
+ "model.layers.36.block_sparse_moe.switch_mlp.gate_proj": true,
464
+ "model.layers.36.block_sparse_moe.switch_mlp.up_proj": true,
465
+ "model.layers.36.block_sparse_moe.switch_mlp.down_proj": true,
466
+ "model.layers.36.block_sparse_moe.switch_mlp.activation": false,
467
+ "model.layers.36.input_layernorm": false,
468
+ "model.layers.36.post_attention_layernorm": false,
469
+ "model.layers.37.self_attn.q_proj": true,
470
+ "model.layers.37.self_attn.k_proj": true,
471
+ "model.layers.37.self_attn.v_proj": true,
472
+ "model.layers.37.self_attn.o_proj": true,
473
+ "model.layers.37.self_attn.rope": false,
474
+ "model.layers.37.block_sparse_moe.gate": true,
475
+ "model.layers.37.block_sparse_moe.switch_mlp.gate_proj": true,
476
+ "model.layers.37.block_sparse_moe.switch_mlp.up_proj": true,
477
+ "model.layers.37.block_sparse_moe.switch_mlp.down_proj": true,
478
+ "model.layers.37.block_sparse_moe.switch_mlp.activation": false,
479
+ "model.layers.37.input_layernorm": false,
480
+ "model.layers.37.post_attention_layernorm": false,
481
+ "model.layers.38.self_attn.q_proj": true,
482
+ "model.layers.38.self_attn.k_proj": true,
483
+ "model.layers.38.self_attn.v_proj": true,
484
+ "model.layers.38.self_attn.o_proj": true,
485
+ "model.layers.38.self_attn.rope": false,
486
+ "model.layers.38.block_sparse_moe.gate": true,
487
+ "model.layers.38.block_sparse_moe.switch_mlp.gate_proj": true,
488
+ "model.layers.38.block_sparse_moe.switch_mlp.up_proj": true,
489
+ "model.layers.38.block_sparse_moe.switch_mlp.down_proj": true,
490
+ "model.layers.38.block_sparse_moe.switch_mlp.activation": false,
491
+ "model.layers.38.input_layernorm": false,
492
+ "model.layers.38.post_attention_layernorm": false,
493
+ "model.layers.39.self_attn.q_proj": true,
494
+ "model.layers.39.self_attn.k_proj": true,
495
+ "model.layers.39.self_attn.v_proj": true,
496
+ "model.layers.39.self_attn.o_proj": true,
497
+ "model.layers.39.self_attn.rope": false,
498
+ "model.layers.39.block_sparse_moe.gate": true,
499
+ "model.layers.39.block_sparse_moe.switch_mlp.gate_proj": true,
500
+ "model.layers.39.block_sparse_moe.switch_mlp.up_proj": true,
501
+ "model.layers.39.block_sparse_moe.switch_mlp.down_proj": true,
502
+ "model.layers.39.block_sparse_moe.switch_mlp.activation": false,
503
+ "model.layers.39.input_layernorm": false,
504
+ "model.layers.39.post_attention_layernorm": false,
505
+ "model.layers.40.self_attn.q_proj": true,
506
+ "model.layers.40.self_attn.k_proj": true,
507
+ "model.layers.40.self_attn.v_proj": true,
508
+ "model.layers.40.self_attn.o_proj": true,
509
+ "model.layers.40.self_attn.rope": false,
510
+ "model.layers.40.block_sparse_moe.gate": true,
511
+ "model.layers.40.block_sparse_moe.switch_mlp.gate_proj": true,
512
+ "model.layers.40.block_sparse_moe.switch_mlp.up_proj": true,
513
+ "model.layers.40.block_sparse_moe.switch_mlp.down_proj": true,
514
+ "model.layers.40.block_sparse_moe.switch_mlp.activation": false,
515
+ "model.layers.40.input_layernorm": false,
516
+ "model.layers.40.post_attention_layernorm": false,
517
+ "model.layers.41.self_attn.q_proj": true,
518
+ "model.layers.41.self_attn.k_proj": true,
519
+ "model.layers.41.self_attn.v_proj": true,
520
+ "model.layers.41.self_attn.o_proj": true,
521
+ "model.layers.41.self_attn.rope": false,
522
+ "model.layers.41.block_sparse_moe.gate": true,
523
+ "model.layers.41.block_sparse_moe.switch_mlp.gate_proj": true,
524
+ "model.layers.41.block_sparse_moe.switch_mlp.up_proj": true,
525
+ "model.layers.41.block_sparse_moe.switch_mlp.down_proj": true,
526
+ "model.layers.41.block_sparse_moe.switch_mlp.activation": false,
527
+ "model.layers.41.input_layernorm": false,
528
+ "model.layers.41.post_attention_layernorm": false,
529
+ "model.layers.42.self_attn.q_proj": true,
530
+ "model.layers.42.self_attn.k_proj": true,
531
+ "model.layers.42.self_attn.v_proj": true,
532
+ "model.layers.42.self_attn.o_proj": true,
533
+ "model.layers.42.self_attn.rope": false,
534
+ "model.layers.42.block_sparse_moe.gate": true,
535
+ "model.layers.42.block_sparse_moe.switch_mlp.gate_proj": true,
536
+ "model.layers.42.block_sparse_moe.switch_mlp.up_proj": true,
537
+ "model.layers.42.block_sparse_moe.switch_mlp.down_proj": true,
538
+ "model.layers.42.block_sparse_moe.switch_mlp.activation": false,
539
+ "model.layers.42.input_layernorm": false,
540
+ "model.layers.42.post_attention_layernorm": false,
541
+ "model.layers.43.self_attn.q_proj": true,
542
+ "model.layers.43.self_attn.k_proj": true,
543
+ "model.layers.43.self_attn.v_proj": true,
544
+ "model.layers.43.self_attn.o_proj": true,
545
+ "model.layers.43.self_attn.rope": false,
546
+ "model.layers.43.block_sparse_moe.gate": true,
547
+ "model.layers.43.block_sparse_moe.switch_mlp.gate_proj": true,
548
+ "model.layers.43.block_sparse_moe.switch_mlp.up_proj": true,
549
+ "model.layers.43.block_sparse_moe.switch_mlp.down_proj": true,
550
+ "model.layers.43.block_sparse_moe.switch_mlp.activation": false,
551
+ "model.layers.43.input_layernorm": false,
552
+ "model.layers.43.post_attention_layernorm": false,
553
+ "model.layers.44.self_attn.q_proj": true,
554
+ "model.layers.44.self_attn.k_proj": true,
555
+ "model.layers.44.self_attn.v_proj": true,
556
+ "model.layers.44.self_attn.o_proj": true,
557
+ "model.layers.44.self_attn.rope": false,
558
+ "model.layers.44.block_sparse_moe.gate": true,
559
+ "model.layers.44.block_sparse_moe.switch_mlp.gate_proj": true,
560
+ "model.layers.44.block_sparse_moe.switch_mlp.up_proj": true,
561
+ "model.layers.44.block_sparse_moe.switch_mlp.down_proj": true,
562
+ "model.layers.44.block_sparse_moe.switch_mlp.activation": false,
563
+ "model.layers.44.input_layernorm": false,
564
+ "model.layers.44.post_attention_layernorm": false,
565
+ "model.layers.45.self_attn.q_proj": true,
566
+ "model.layers.45.self_attn.k_proj": true,
567
+ "model.layers.45.self_attn.v_proj": true,
568
+ "model.layers.45.self_attn.o_proj": true,
569
+ "model.layers.45.self_attn.rope": false,
570
+ "model.layers.45.block_sparse_moe.gate": true,
571
+ "model.layers.45.block_sparse_moe.switch_mlp.gate_proj": true,
572
+ "model.layers.45.block_sparse_moe.switch_mlp.up_proj": true,
573
+ "model.layers.45.block_sparse_moe.switch_mlp.down_proj": true,
574
+ "model.layers.45.block_sparse_moe.switch_mlp.activation": false,
575
+ "model.layers.45.input_layernorm": false,
576
+ "model.layers.45.post_attention_layernorm": false,
577
+ "model.layers.46.self_attn.q_proj": true,
578
+ "model.layers.46.self_attn.k_proj": true,
579
+ "model.layers.46.self_attn.v_proj": true,
580
+ "model.layers.46.self_attn.o_proj": true,
581
+ "model.layers.46.self_attn.rope": false,
582
+ "model.layers.46.block_sparse_moe.gate": true,
583
+ "model.layers.46.block_sparse_moe.switch_mlp.gate_proj": true,
584
+ "model.layers.46.block_sparse_moe.switch_mlp.up_proj": true,
585
+ "model.layers.46.block_sparse_moe.switch_mlp.down_proj": true,
586
+ "model.layers.46.block_sparse_moe.switch_mlp.activation": false,
587
+ "model.layers.46.input_layernorm": false,
588
+ "model.layers.46.post_attention_layernorm": false,
589
+ "model.layers.47.self_attn.q_proj": true,
590
+ "model.layers.47.self_attn.k_proj": true,
591
+ "model.layers.47.self_attn.v_proj": true,
592
+ "model.layers.47.self_attn.o_proj": true,
593
+ "model.layers.47.self_attn.rope": false,
594
+ "model.layers.47.block_sparse_moe.gate": true,
595
+ "model.layers.47.block_sparse_moe.switch_mlp.gate_proj": true,
596
+ "model.layers.47.block_sparse_moe.switch_mlp.up_proj": true,
597
+ "model.layers.47.block_sparse_moe.switch_mlp.down_proj": true,
598
+ "model.layers.47.block_sparse_moe.switch_mlp.activation": false,
599
+ "model.layers.47.input_layernorm": false,
600
+ "model.layers.47.post_attention_layernorm": false,
601
+ "model.layers.48.self_attn.q_proj": true,
602
+ "model.layers.48.self_attn.k_proj": true,
603
+ "model.layers.48.self_attn.v_proj": true,
604
+ "model.layers.48.self_attn.o_proj": true,
605
+ "model.layers.48.self_attn.rope": false,
606
+ "model.layers.48.block_sparse_moe.gate": true,
607
+ "model.layers.48.block_sparse_moe.switch_mlp.gate_proj": true,
608
+ "model.layers.48.block_sparse_moe.switch_mlp.up_proj": true,
609
+ "model.layers.48.block_sparse_moe.switch_mlp.down_proj": true,
610
+ "model.layers.48.block_sparse_moe.switch_mlp.activation": false,
611
+ "model.layers.48.input_layernorm": false,
612
+ "model.layers.48.post_attention_layernorm": false,
613
+ "model.layers.49.self_attn.q_proj": true,
614
+ "model.layers.49.self_attn.k_proj": true,
615
+ "model.layers.49.self_attn.v_proj": true,
616
+ "model.layers.49.self_attn.o_proj": true,
617
+ "model.layers.49.self_attn.rope": false,
618
+ "model.layers.49.block_sparse_moe.gate": true,
619
+ "model.layers.49.block_sparse_moe.switch_mlp.gate_proj": true,
620
+ "model.layers.49.block_sparse_moe.switch_mlp.up_proj": true,
621
+ "model.layers.49.block_sparse_moe.switch_mlp.down_proj": true,
622
+ "model.layers.49.block_sparse_moe.switch_mlp.activation": false,
623
+ "model.layers.49.input_layernorm": false,
624
+ "model.layers.49.post_attention_layernorm": false,
625
+ "model.layers.50.self_attn.q_proj": true,
626
+ "model.layers.50.self_attn.k_proj": true,
627
+ "model.layers.50.self_attn.v_proj": true,
628
+ "model.layers.50.self_attn.o_proj": true,
629
+ "model.layers.50.self_attn.rope": false,
630
+ "model.layers.50.block_sparse_moe.gate": true,
631
+ "model.layers.50.block_sparse_moe.switch_mlp.gate_proj": true,
632
+ "model.layers.50.block_sparse_moe.switch_mlp.up_proj": true,
633
+ "model.layers.50.block_sparse_moe.switch_mlp.down_proj": true,
634
+ "model.layers.50.block_sparse_moe.switch_mlp.activation": false,
635
+ "model.layers.50.input_layernorm": false,
636
+ "model.layers.50.post_attention_layernorm": false,
637
+ "model.layers.51.self_attn.q_proj": true,
638
+ "model.layers.51.self_attn.k_proj": true,
639
+ "model.layers.51.self_attn.v_proj": true,
640
+ "model.layers.51.self_attn.o_proj": true,
641
+ "model.layers.51.self_attn.rope": false,
642
+ "model.layers.51.block_sparse_moe.gate": true,
643
+ "model.layers.51.block_sparse_moe.switch_mlp.gate_proj": true,
644
+ "model.layers.51.block_sparse_moe.switch_mlp.up_proj": true,
645
+ "model.layers.51.block_sparse_moe.switch_mlp.down_proj": true,
646
+ "model.layers.51.block_sparse_moe.switch_mlp.activation": false,
647
+ "model.layers.51.input_layernorm": false,
648
+ "model.layers.51.post_attention_layernorm": false,
649
+ "model.layers.52.self_attn.q_proj": true,
650
+ "model.layers.52.self_attn.k_proj": true,
651
+ "model.layers.52.self_attn.v_proj": true,
652
+ "model.layers.52.self_attn.o_proj": true,
653
+ "model.layers.52.self_attn.rope": false,
654
+ "model.layers.52.block_sparse_moe.gate": true,
655
+ "model.layers.52.block_sparse_moe.switch_mlp.gate_proj": true,
656
+ "model.layers.52.block_sparse_moe.switch_mlp.up_proj": true,
657
+ "model.layers.52.block_sparse_moe.switch_mlp.down_proj": true,
658
+ "model.layers.52.block_sparse_moe.switch_mlp.activation": false,
659
+ "model.layers.52.input_layernorm": false,
660
+ "model.layers.52.post_attention_layernorm": false,
661
+ "model.layers.53.self_attn.q_proj": true,
662
+ "model.layers.53.self_attn.k_proj": true,
663
+ "model.layers.53.self_attn.v_proj": true,
664
+ "model.layers.53.self_attn.o_proj": true,
665
+ "model.layers.53.self_attn.rope": false,
666
+ "model.layers.53.block_sparse_moe.gate": true,
667
+ "model.layers.53.block_sparse_moe.switch_mlp.gate_proj": true,
668
+ "model.layers.53.block_sparse_moe.switch_mlp.up_proj": true,
669
+ "model.layers.53.block_sparse_moe.switch_mlp.down_proj": true,
670
+ "model.layers.53.block_sparse_moe.switch_mlp.activation": false,
671
+ "model.layers.53.input_layernorm": false,
672
+ "model.layers.53.post_attention_layernorm": false,
673
+ "model.layers.54.self_attn.q_proj": true,
674
+ "model.layers.54.self_attn.k_proj": true,
675
+ "model.layers.54.self_attn.v_proj": true,
676
+ "model.layers.54.self_attn.o_proj": true,
677
+ "model.layers.54.self_attn.rope": false,
678
+ "model.layers.54.block_sparse_moe.gate": true,
679
+ "model.layers.54.block_sparse_moe.switch_mlp.gate_proj": true,
680
+ "model.layers.54.block_sparse_moe.switch_mlp.up_proj": true,
681
+ "model.layers.54.block_sparse_moe.switch_mlp.down_proj": true,
682
+ "model.layers.54.block_sparse_moe.switch_mlp.activation": false,
683
+ "model.layers.54.input_layernorm": false,
684
+ "model.layers.54.post_attention_layernorm": false,
685
+ "model.layers.55.self_attn.q_proj": true,
686
+ "model.layers.55.self_attn.k_proj": true,
687
+ "model.layers.55.self_attn.v_proj": true,
688
+ "model.layers.55.self_attn.o_proj": true,
689
+ "model.layers.55.self_attn.rope": false,
690
+ "model.layers.55.block_sparse_moe.gate": true,
691
+ "model.layers.55.block_sparse_moe.switch_mlp.gate_proj": true,
692
+ "model.layers.55.block_sparse_moe.switch_mlp.up_proj": true,
693
+ "model.layers.55.block_sparse_moe.switch_mlp.down_proj": true,
694
+ "model.layers.55.block_sparse_moe.switch_mlp.activation": false,
695
+ "model.layers.55.input_layernorm": false,
696
+ "model.layers.55.post_attention_layernorm": false,
697
+ "model.norm": false,
698
+ "lm_head": true
699
+ },
700
+ "quantization_config": {
701
+ "group_size": 64,
702
+ "bits": 4,
703
+ "model.embed_tokens": true,
704
+ "model.layers.0.self_attn.q_proj": true,
705
+ "model.layers.0.self_attn.k_proj": true,
706
+ "model.layers.0.self_attn.v_proj": true,
707
+ "model.layers.0.self_attn.o_proj": true,
708
+ "model.layers.0.self_attn.rope": false,
709
+ "model.layers.0.block_sparse_moe.gate": true,
710
+ "model.layers.0.block_sparse_moe.switch_mlp.gate_proj": true,
711
+ "model.layers.0.block_sparse_moe.switch_mlp.up_proj": true,
712
+ "model.layers.0.block_sparse_moe.switch_mlp.down_proj": true,
713
+ "model.layers.0.block_sparse_moe.switch_mlp.activation": false,
714
+ "model.layers.0.input_layernorm": false,
715
+ "model.layers.0.post_attention_layernorm": false,
716
+ "model.layers.1.self_attn.q_proj": true,
717
+ "model.layers.1.self_attn.k_proj": true,
718
+ "model.layers.1.self_attn.v_proj": true,
719
+ "model.layers.1.self_attn.o_proj": true,
720
+ "model.layers.1.self_attn.rope": false,
721
+ "model.layers.1.block_sparse_moe.gate": true,
722
+ "model.layers.1.block_sparse_moe.switch_mlp.gate_proj": true,
723
+ "model.layers.1.block_sparse_moe.switch_mlp.up_proj": true,
724
+ "model.layers.1.block_sparse_moe.switch_mlp.down_proj": true,
725
+ "model.layers.1.block_sparse_moe.switch_mlp.activation": false,
726
+ "model.layers.1.input_layernorm": false,
727
+ "model.layers.1.post_attention_layernorm": false,
728
+ "model.layers.2.self_attn.q_proj": true,
729
+ "model.layers.2.self_attn.k_proj": true,
730
+ "model.layers.2.self_attn.v_proj": true,
731
+ "model.layers.2.self_attn.o_proj": true,
732
+ "model.layers.2.self_attn.rope": false,
733
+ "model.layers.2.block_sparse_moe.gate": true,
734
+ "model.layers.2.block_sparse_moe.switch_mlp.gate_proj": true,
735
+ "model.layers.2.block_sparse_moe.switch_mlp.up_proj": true,
736
+ "model.layers.2.block_sparse_moe.switch_mlp.down_proj": true,
737
+ "model.layers.2.block_sparse_moe.switch_mlp.activation": false,
738
+ "model.layers.2.input_layernorm": false,
739
+ "model.layers.2.post_attention_layernorm": false,
740
+ "model.layers.3.self_attn.q_proj": true,
741
+ "model.layers.3.self_attn.k_proj": true,
742
+ "model.layers.3.self_attn.v_proj": true,
743
+ "model.layers.3.self_attn.o_proj": true,
744
+ "model.layers.3.self_attn.rope": false,
745
+ "model.layers.3.block_sparse_moe.gate": true,
746
+ "model.layers.3.block_sparse_moe.switch_mlp.gate_proj": true,
747
+ "model.layers.3.block_sparse_moe.switch_mlp.up_proj": true,
748
+ "model.layers.3.block_sparse_moe.switch_mlp.down_proj": true,
749
+ "model.layers.3.block_sparse_moe.switch_mlp.activation": false,
750
+ "model.layers.3.input_layernorm": false,
751
+ "model.layers.3.post_attention_layernorm": false,
752
+ "model.layers.4.self_attn.q_proj": true,
753
+ "model.layers.4.self_attn.k_proj": true,
754
+ "model.layers.4.self_attn.v_proj": true,
755
+ "model.layers.4.self_attn.o_proj": true,
756
+ "model.layers.4.self_attn.rope": false,
757
+ "model.layers.4.block_sparse_moe.gate": true,
758
+ "model.layers.4.block_sparse_moe.switch_mlp.gate_proj": true,
759
+ "model.layers.4.block_sparse_moe.switch_mlp.up_proj": true,
760
+ "model.layers.4.block_sparse_moe.switch_mlp.down_proj": true,
761
+ "model.layers.4.block_sparse_moe.switch_mlp.activation": false,
762
+ "model.layers.4.input_layernorm": false,
763
+ "model.layers.4.post_attention_layernorm": false,
764
+ "model.layers.5.self_attn.q_proj": true,
765
+ "model.layers.5.self_attn.k_proj": true,
766
+ "model.layers.5.self_attn.v_proj": true,
767
+ "model.layers.5.self_attn.o_proj": true,
768
+ "model.layers.5.self_attn.rope": false,
769
+ "model.layers.5.block_sparse_moe.gate": true,
770
+ "model.layers.5.block_sparse_moe.switch_mlp.gate_proj": true,
771
+ "model.layers.5.block_sparse_moe.switch_mlp.up_proj": true,
772
+ "model.layers.5.block_sparse_moe.switch_mlp.down_proj": true,
773
+ "model.layers.5.block_sparse_moe.switch_mlp.activation": false,
774
+ "model.layers.5.input_layernorm": false,
775
+ "model.layers.5.post_attention_layernorm": false,
776
+ "model.layers.6.self_attn.q_proj": true,
777
+ "model.layers.6.self_attn.k_proj": true,
778
+ "model.layers.6.self_attn.v_proj": true,
779
+ "model.layers.6.self_attn.o_proj": true,
780
+ "model.layers.6.self_attn.rope": false,
781
+ "model.layers.6.block_sparse_moe.gate": true,
782
+ "model.layers.6.block_sparse_moe.switch_mlp.gate_proj": true,
783
+ "model.layers.6.block_sparse_moe.switch_mlp.up_proj": true,
784
+ "model.layers.6.block_sparse_moe.switch_mlp.down_proj": true,
785
+ "model.layers.6.block_sparse_moe.switch_mlp.activation": false,
786
+ "model.layers.6.input_layernorm": false,
787
+ "model.layers.6.post_attention_layernorm": false,
788
+ "model.layers.7.self_attn.q_proj": true,
789
+ "model.layers.7.self_attn.k_proj": true,
790
+ "model.layers.7.self_attn.v_proj": true,
791
+ "model.layers.7.self_attn.o_proj": true,
792
+ "model.layers.7.self_attn.rope": false,
793
+ "model.layers.7.block_sparse_moe.gate": true,
794
+ "model.layers.7.block_sparse_moe.switch_mlp.gate_proj": true,
795
+ "model.layers.7.block_sparse_moe.switch_mlp.up_proj": true,
796
+ "model.layers.7.block_sparse_moe.switch_mlp.down_proj": true,
797
+ "model.layers.7.block_sparse_moe.switch_mlp.activation": false,
798
+ "model.layers.7.input_layernorm": false,
799
+ "model.layers.7.post_attention_layernorm": false,
800
+ "model.layers.8.self_attn.q_proj": true,
801
+ "model.layers.8.self_attn.k_proj": true,
802
+ "model.layers.8.self_attn.v_proj": true,
803
+ "model.layers.8.self_attn.o_proj": true,
804
+ "model.layers.8.self_attn.rope": false,
805
+ "model.layers.8.block_sparse_moe.gate": true,
806
+ "model.layers.8.block_sparse_moe.switch_mlp.gate_proj": true,
807
+ "model.layers.8.block_sparse_moe.switch_mlp.up_proj": true,
808
+ "model.layers.8.block_sparse_moe.switch_mlp.down_proj": true,
809
+ "model.layers.8.block_sparse_moe.switch_mlp.activation": false,
810
+ "model.layers.8.input_layernorm": false,
811
+ "model.layers.8.post_attention_layernorm": false,
812
+ "model.layers.9.self_attn.q_proj": true,
813
+ "model.layers.9.self_attn.k_proj": true,
814
+ "model.layers.9.self_attn.v_proj": true,
815
+ "model.layers.9.self_attn.o_proj": true,
816
+ "model.layers.9.self_attn.rope": false,
817
+ "model.layers.9.block_sparse_moe.gate": true,
818
+ "model.layers.9.block_sparse_moe.switch_mlp.gate_proj": true,
819
+ "model.layers.9.block_sparse_moe.switch_mlp.up_proj": true,
820
+ "model.layers.9.block_sparse_moe.switch_mlp.down_proj": true,
821
+ "model.layers.9.block_sparse_moe.switch_mlp.activation": false,
822
+ "model.layers.9.input_layernorm": false,
823
+ "model.layers.9.post_attention_layernorm": false,
824
+ "model.layers.10.self_attn.q_proj": true,
825
+ "model.layers.10.self_attn.k_proj": true,
826
+ "model.layers.10.self_attn.v_proj": true,
827
+ "model.layers.10.self_attn.o_proj": true,
828
+ "model.layers.10.self_attn.rope": false,
829
+ "model.layers.10.block_sparse_moe.gate": true,
830
+ "model.layers.10.block_sparse_moe.switch_mlp.gate_proj": true,
831
+ "model.layers.10.block_sparse_moe.switch_mlp.up_proj": true,
832
+ "model.layers.10.block_sparse_moe.switch_mlp.down_proj": true,
833
+ "model.layers.10.block_sparse_moe.switch_mlp.activation": false,
834
+ "model.layers.10.input_layernorm": false,
835
+ "model.layers.10.post_attention_layernorm": false,
836
+ "model.layers.11.self_attn.q_proj": true,
837
+ "model.layers.11.self_attn.k_proj": true,
838
+ "model.layers.11.self_attn.v_proj": true,
839
+ "model.layers.11.self_attn.o_proj": true,
840
+ "model.layers.11.self_attn.rope": false,
841
+ "model.layers.11.block_sparse_moe.gate": true,
842
+ "model.layers.11.block_sparse_moe.switch_mlp.gate_proj": true,
843
+ "model.layers.11.block_sparse_moe.switch_mlp.up_proj": true,
844
+ "model.layers.11.block_sparse_moe.switch_mlp.down_proj": true,
845
+ "model.layers.11.block_sparse_moe.switch_mlp.activation": false,
846
+ "model.layers.11.input_layernorm": false,
847
+ "model.layers.11.post_attention_layernorm": false,
848
+ "model.layers.12.self_attn.q_proj": true,
849
+ "model.layers.12.self_attn.k_proj": true,
850
+ "model.layers.12.self_attn.v_proj": true,
851
+ "model.layers.12.self_attn.o_proj": true,
852
+ "model.layers.12.self_attn.rope": false,
853
+ "model.layers.12.block_sparse_moe.gate": true,
854
+ "model.layers.12.block_sparse_moe.switch_mlp.gate_proj": true,
855
+ "model.layers.12.block_sparse_moe.switch_mlp.up_proj": true,
856
+ "model.layers.12.block_sparse_moe.switch_mlp.down_proj": true,
857
+ "model.layers.12.block_sparse_moe.switch_mlp.activation": false,
858
+ "model.layers.12.input_layernorm": false,
859
+ "model.layers.12.post_attention_layernorm": false,
860
+ "model.layers.13.self_attn.q_proj": true,
861
+ "model.layers.13.self_attn.k_proj": true,
862
+ "model.layers.13.self_attn.v_proj": true,
863
+ "model.layers.13.self_attn.o_proj": true,
864
+ "model.layers.13.self_attn.rope": false,
865
+ "model.layers.13.block_sparse_moe.gate": true,
866
+ "model.layers.13.block_sparse_moe.switch_mlp.gate_proj": true,
867
+ "model.layers.13.block_sparse_moe.switch_mlp.up_proj": true,
868
+ "model.layers.13.block_sparse_moe.switch_mlp.down_proj": true,
869
+ "model.layers.13.block_sparse_moe.switch_mlp.activation": false,
870
+ "model.layers.13.input_layernorm": false,
871
+ "model.layers.13.post_attention_layernorm": false,
872
+ "model.layers.14.self_attn.q_proj": true,
873
+ "model.layers.14.self_attn.k_proj": true,
874
+ "model.layers.14.self_attn.v_proj": true,
875
+ "model.layers.14.self_attn.o_proj": true,
876
+ "model.layers.14.self_attn.rope": false,
877
+ "model.layers.14.block_sparse_moe.gate": true,
878
+ "model.layers.14.block_sparse_moe.switch_mlp.gate_proj": true,
879
+ "model.layers.14.block_sparse_moe.switch_mlp.up_proj": true,
880
+ "model.layers.14.block_sparse_moe.switch_mlp.down_proj": true,
881
+ "model.layers.14.block_sparse_moe.switch_mlp.activation": false,
882
+ "model.layers.14.input_layernorm": false,
883
+ "model.layers.14.post_attention_layernorm": false,
884
+ "model.layers.15.self_attn.q_proj": true,
885
+ "model.layers.15.self_attn.k_proj": true,
886
+ "model.layers.15.self_attn.v_proj": true,
887
+ "model.layers.15.self_attn.o_proj": true,
888
+ "model.layers.15.self_attn.rope": false,
889
+ "model.layers.15.block_sparse_moe.gate": true,
890
+ "model.layers.15.block_sparse_moe.switch_mlp.gate_proj": true,
891
+ "model.layers.15.block_sparse_moe.switch_mlp.up_proj": true,
892
+ "model.layers.15.block_sparse_moe.switch_mlp.down_proj": true,
893
+ "model.layers.15.block_sparse_moe.switch_mlp.activation": false,
894
+ "model.layers.15.input_layernorm": false,
895
+ "model.layers.15.post_attention_layernorm": false,
896
+ "model.layers.16.self_attn.q_proj": true,
897
+ "model.layers.16.self_attn.k_proj": true,
898
+ "model.layers.16.self_attn.v_proj": true,
899
+ "model.layers.16.self_attn.o_proj": true,
900
+ "model.layers.16.self_attn.rope": false,
901
+ "model.layers.16.block_sparse_moe.gate": true,
902
+ "model.layers.16.block_sparse_moe.switch_mlp.gate_proj": true,
903
+ "model.layers.16.block_sparse_moe.switch_mlp.up_proj": true,
904
+ "model.layers.16.block_sparse_moe.switch_mlp.down_proj": true,
905
+ "model.layers.16.block_sparse_moe.switch_mlp.activation": false,
906
+ "model.layers.16.input_layernorm": false,
907
+ "model.layers.16.post_attention_layernorm": false,
908
+ "model.layers.17.self_attn.q_proj": true,
909
+ "model.layers.17.self_attn.k_proj": true,
910
+ "model.layers.17.self_attn.v_proj": true,
911
+ "model.layers.17.self_attn.o_proj": true,
912
+ "model.layers.17.self_attn.rope": false,
913
+ "model.layers.17.block_sparse_moe.gate": true,
914
+ "model.layers.17.block_sparse_moe.switch_mlp.gate_proj": true,
915
+ "model.layers.17.block_sparse_moe.switch_mlp.up_proj": true,
916
+ "model.layers.17.block_sparse_moe.switch_mlp.down_proj": true,
917
+ "model.layers.17.block_sparse_moe.switch_mlp.activation": false,
918
+ "model.layers.17.input_layernorm": false,
919
+ "model.layers.17.post_attention_layernorm": false,
920
+ "model.layers.18.self_attn.q_proj": true,
921
+ "model.layers.18.self_attn.k_proj": true,
922
+ "model.layers.18.self_attn.v_proj": true,
923
+ "model.layers.18.self_attn.o_proj": true,
924
+ "model.layers.18.self_attn.rope": false,
925
+ "model.layers.18.block_sparse_moe.gate": true,
926
+ "model.layers.18.block_sparse_moe.switch_mlp.gate_proj": true,
927
+ "model.layers.18.block_sparse_moe.switch_mlp.up_proj": true,
928
+ "model.layers.18.block_sparse_moe.switch_mlp.down_proj": true,
929
+ "model.layers.18.block_sparse_moe.switch_mlp.activation": false,
930
+ "model.layers.18.input_layernorm": false,
931
+ "model.layers.18.post_attention_layernorm": false,
932
+ "model.layers.19.self_attn.q_proj": true,
933
+ "model.layers.19.self_attn.k_proj": true,
934
+ "model.layers.19.self_attn.v_proj": true,
935
+ "model.layers.19.self_attn.o_proj": true,
936
+ "model.layers.19.self_attn.rope": false,
937
+ "model.layers.19.block_sparse_moe.gate": true,
938
+ "model.layers.19.block_sparse_moe.switch_mlp.gate_proj": true,
939
+ "model.layers.19.block_sparse_moe.switch_mlp.up_proj": true,
940
+ "model.layers.19.block_sparse_moe.switch_mlp.down_proj": true,
941
+ "model.layers.19.block_sparse_moe.switch_mlp.activation": false,
942
+ "model.layers.19.input_layernorm": false,
943
+ "model.layers.19.post_attention_layernorm": false,
944
+ "model.layers.20.self_attn.q_proj": true,
945
+ "model.layers.20.self_attn.k_proj": true,
946
+ "model.layers.20.self_attn.v_proj": true,
947
+ "model.layers.20.self_attn.o_proj": true,
948
+ "model.layers.20.self_attn.rope": false,
949
+ "model.layers.20.block_sparse_moe.gate": true,
950
+ "model.layers.20.block_sparse_moe.switch_mlp.gate_proj": true,
951
+ "model.layers.20.block_sparse_moe.switch_mlp.up_proj": true,
952
+ "model.layers.20.block_sparse_moe.switch_mlp.down_proj": true,
953
+ "model.layers.20.block_sparse_moe.switch_mlp.activation": false,
954
+ "model.layers.20.input_layernorm": false,
955
+ "model.layers.20.post_attention_layernorm": false,
956
+ "model.layers.21.self_attn.q_proj": true,
957
+ "model.layers.21.self_attn.k_proj": true,
958
+ "model.layers.21.self_attn.v_proj": true,
959
+ "model.layers.21.self_attn.o_proj": true,
960
+ "model.layers.21.self_attn.rope": false,
961
+ "model.layers.21.block_sparse_moe.gate": true,
962
+ "model.layers.21.block_sparse_moe.switch_mlp.gate_proj": true,
963
+ "model.layers.21.block_sparse_moe.switch_mlp.up_proj": true,
964
+ "model.layers.21.block_sparse_moe.switch_mlp.down_proj": true,
965
+ "model.layers.21.block_sparse_moe.switch_mlp.activation": false,
966
+ "model.layers.21.input_layernorm": false,
967
+ "model.layers.21.post_attention_layernorm": false,
968
+ "model.layers.22.self_attn.q_proj": true,
969
+ "model.layers.22.self_attn.k_proj": true,
970
+ "model.layers.22.self_attn.v_proj": true,
971
+ "model.layers.22.self_attn.o_proj": true,
972
+ "model.layers.22.self_attn.rope": false,
973
+ "model.layers.22.block_sparse_moe.gate": true,
974
+ "model.layers.22.block_sparse_moe.switch_mlp.gate_proj": true,
975
+ "model.layers.22.block_sparse_moe.switch_mlp.up_proj": true,
976
+ "model.layers.22.block_sparse_moe.switch_mlp.down_proj": true,
977
+ "model.layers.22.block_sparse_moe.switch_mlp.activation": false,
978
+ "model.layers.22.input_layernorm": false,
979
+ "model.layers.22.post_attention_layernorm": false,
980
+ "model.layers.23.self_attn.q_proj": true,
981
+ "model.layers.23.self_attn.k_proj": true,
982
+ "model.layers.23.self_attn.v_proj": true,
983
+ "model.layers.23.self_attn.o_proj": true,
984
+ "model.layers.23.self_attn.rope": false,
985
+ "model.layers.23.block_sparse_moe.gate": true,
986
+ "model.layers.23.block_sparse_moe.switch_mlp.gate_proj": true,
987
+ "model.layers.23.block_sparse_moe.switch_mlp.up_proj": true,
988
+ "model.layers.23.block_sparse_moe.switch_mlp.down_proj": true,
989
+ "model.layers.23.block_sparse_moe.switch_mlp.activation": false,
990
+ "model.layers.23.input_layernorm": false,
991
+ "model.layers.23.post_attention_layernorm": false,
992
+ "model.layers.24.self_attn.q_proj": true,
993
+ "model.layers.24.self_attn.k_proj": true,
994
+ "model.layers.24.self_attn.v_proj": true,
995
+ "model.layers.24.self_attn.o_proj": true,
996
+ "model.layers.24.self_attn.rope": false,
997
+ "model.layers.24.block_sparse_moe.gate": true,
998
+ "model.layers.24.block_sparse_moe.switch_mlp.gate_proj": true,
999
+ "model.layers.24.block_sparse_moe.switch_mlp.up_proj": true,
1000
+ "model.layers.24.block_sparse_moe.switch_mlp.down_proj": true,
1001
+ "model.layers.24.block_sparse_moe.switch_mlp.activation": false,
1002
+ "model.layers.24.input_layernorm": false,
1003
+ "model.layers.24.post_attention_layernorm": false,
1004
+ "model.layers.25.self_attn.q_proj": true,
1005
+ "model.layers.25.self_attn.k_proj": true,
1006
+ "model.layers.25.self_attn.v_proj": true,
1007
+ "model.layers.25.self_attn.o_proj": true,
1008
+ "model.layers.25.self_attn.rope": false,
1009
+ "model.layers.25.block_sparse_moe.gate": true,
1010
+ "model.layers.25.block_sparse_moe.switch_mlp.gate_proj": true,
1011
+ "model.layers.25.block_sparse_moe.switch_mlp.up_proj": true,
1012
+ "model.layers.25.block_sparse_moe.switch_mlp.down_proj": true,
1013
+ "model.layers.25.block_sparse_moe.switch_mlp.activation": false,
1014
+ "model.layers.25.input_layernorm": false,
1015
+ "model.layers.25.post_attention_layernorm": false,
1016
+ "model.layers.26.self_attn.q_proj": true,
1017
+ "model.layers.26.self_attn.k_proj": true,
1018
+ "model.layers.26.self_attn.v_proj": true,
1019
+ "model.layers.26.self_attn.o_proj": true,
1020
+ "model.layers.26.self_attn.rope": false,
1021
+ "model.layers.26.block_sparse_moe.gate": true,
1022
+ "model.layers.26.block_sparse_moe.switch_mlp.gate_proj": true,
1023
+ "model.layers.26.block_sparse_moe.switch_mlp.up_proj": true,
1024
+ "model.layers.26.block_sparse_moe.switch_mlp.down_proj": true,
1025
+ "model.layers.26.block_sparse_moe.switch_mlp.activation": false,
1026
+ "model.layers.26.input_layernorm": false,
1027
+ "model.layers.26.post_attention_layernorm": false,
1028
+ "model.layers.27.self_attn.q_proj": true,
1029
+ "model.layers.27.self_attn.k_proj": true,
1030
+ "model.layers.27.self_attn.v_proj": true,
1031
+ "model.layers.27.self_attn.o_proj": true,
1032
+ "model.layers.27.self_attn.rope": false,
1033
+ "model.layers.27.block_sparse_moe.gate": true,
1034
+ "model.layers.27.block_sparse_moe.switch_mlp.gate_proj": true,
1035
+ "model.layers.27.block_sparse_moe.switch_mlp.up_proj": true,
1036
+ "model.layers.27.block_sparse_moe.switch_mlp.down_proj": true,
1037
+ "model.layers.27.block_sparse_moe.switch_mlp.activation": false,
1038
+ "model.layers.27.input_layernorm": false,
1039
+ "model.layers.27.post_attention_layernorm": false,
1040
+ "model.layers.28.self_attn.q_proj": true,
1041
+ "model.layers.28.self_attn.k_proj": true,
1042
+ "model.layers.28.self_attn.v_proj": true,
1043
+ "model.layers.28.self_attn.o_proj": true,
1044
+ "model.layers.28.self_attn.rope": false,
1045
+ "model.layers.28.block_sparse_moe.gate": true,
1046
+ "model.layers.28.block_sparse_moe.switch_mlp.gate_proj": true,
1047
+ "model.layers.28.block_sparse_moe.switch_mlp.up_proj": true,
1048
+ "model.layers.28.block_sparse_moe.switch_mlp.down_proj": true,
1049
+ "model.layers.28.block_sparse_moe.switch_mlp.activation": false,
1050
+ "model.layers.28.input_layernorm": false,
1051
+ "model.layers.28.post_attention_layernorm": false,
1052
+ "model.layers.29.self_attn.q_proj": true,
1053
+ "model.layers.29.self_attn.k_proj": true,
1054
+ "model.layers.29.self_attn.v_proj": true,
1055
+ "model.layers.29.self_attn.o_proj": true,
1056
+ "model.layers.29.self_attn.rope": false,
1057
+ "model.layers.29.block_sparse_moe.gate": true,
1058
+ "model.layers.29.block_sparse_moe.switch_mlp.gate_proj": true,
1059
+ "model.layers.29.block_sparse_moe.switch_mlp.up_proj": true,
1060
+ "model.layers.29.block_sparse_moe.switch_mlp.down_proj": true,
1061
+ "model.layers.29.block_sparse_moe.switch_mlp.activation": false,
1062
+ "model.layers.29.input_layernorm": false,
1063
+ "model.layers.29.post_attention_layernorm": false,
1064
+ "model.layers.30.self_attn.q_proj": true,
1065
+ "model.layers.30.self_attn.k_proj": true,
1066
+ "model.layers.30.self_attn.v_proj": true,
1067
+ "model.layers.30.self_attn.o_proj": true,
1068
+ "model.layers.30.self_attn.rope": false,
1069
+ "model.layers.30.block_sparse_moe.gate": true,
1070
+ "model.layers.30.block_sparse_moe.switch_mlp.gate_proj": true,
1071
+ "model.layers.30.block_sparse_moe.switch_mlp.up_proj": true,
1072
+ "model.layers.30.block_sparse_moe.switch_mlp.down_proj": true,
1073
+ "model.layers.30.block_sparse_moe.switch_mlp.activation": false,
1074
+ "model.layers.30.input_layernorm": false,
1075
+ "model.layers.30.post_attention_layernorm": false,
1076
+ "model.layers.31.self_attn.q_proj": true,
1077
+ "model.layers.31.self_attn.k_proj": true,
1078
+ "model.layers.31.self_attn.v_proj": true,
1079
+ "model.layers.31.self_attn.o_proj": true,
1080
+ "model.layers.31.self_attn.rope": false,
1081
+ "model.layers.31.block_sparse_moe.gate": true,
1082
+ "model.layers.31.block_sparse_moe.switch_mlp.gate_proj": true,
1083
+ "model.layers.31.block_sparse_moe.switch_mlp.up_proj": true,
1084
+ "model.layers.31.block_sparse_moe.switch_mlp.down_proj": true,
1085
+ "model.layers.31.block_sparse_moe.switch_mlp.activation": false,
1086
+ "model.layers.31.input_layernorm": false,
1087
+ "model.layers.31.post_attention_layernorm": false,
1088
+ "model.layers.32.self_attn.q_proj": true,
1089
+ "model.layers.32.self_attn.k_proj": true,
1090
+ "model.layers.32.self_attn.v_proj": true,
1091
+ "model.layers.32.self_attn.o_proj": true,
1092
+ "model.layers.32.self_attn.rope": false,
1093
+ "model.layers.32.block_sparse_moe.gate": true,
1094
+ "model.layers.32.block_sparse_moe.switch_mlp.gate_proj": true,
1095
+ "model.layers.32.block_sparse_moe.switch_mlp.up_proj": true,
1096
+ "model.layers.32.block_sparse_moe.switch_mlp.down_proj": true,
1097
+ "model.layers.32.block_sparse_moe.switch_mlp.activation": false,
1098
+ "model.layers.32.input_layernorm": false,
1099
+ "model.layers.32.post_attention_layernorm": false,
1100
+ "model.layers.33.self_attn.q_proj": true,
1101
+ "model.layers.33.self_attn.k_proj": true,
1102
+ "model.layers.33.self_attn.v_proj": true,
1103
+ "model.layers.33.self_attn.o_proj": true,
1104
+ "model.layers.33.self_attn.rope": false,
1105
+ "model.layers.33.block_sparse_moe.gate": true,
1106
+ "model.layers.33.block_sparse_moe.switch_mlp.gate_proj": true,
1107
+ "model.layers.33.block_sparse_moe.switch_mlp.up_proj": true,
1108
+ "model.layers.33.block_sparse_moe.switch_mlp.down_proj": true,
1109
+ "model.layers.33.block_sparse_moe.switch_mlp.activation": false,
1110
+ "model.layers.33.input_layernorm": false,
1111
+ "model.layers.33.post_attention_layernorm": false,
1112
+ "model.layers.34.self_attn.q_proj": true,
1113
+ "model.layers.34.self_attn.k_proj": true,
1114
+ "model.layers.34.self_attn.v_proj": true,
1115
+ "model.layers.34.self_attn.o_proj": true,
1116
+ "model.layers.34.self_attn.rope": false,
1117
+ "model.layers.34.block_sparse_moe.gate": true,
1118
+ "model.layers.34.block_sparse_moe.switch_mlp.gate_proj": true,
1119
+ "model.layers.34.block_sparse_moe.switch_mlp.up_proj": true,
1120
+ "model.layers.34.block_sparse_moe.switch_mlp.down_proj": true,
1121
+ "model.layers.34.block_sparse_moe.switch_mlp.activation": false,
1122
+ "model.layers.34.input_layernorm": false,
1123
+ "model.layers.34.post_attention_layernorm": false,
1124
+ "model.layers.35.self_attn.q_proj": true,
1125
+ "model.layers.35.self_attn.k_proj": true,
1126
+ "model.layers.35.self_attn.v_proj": true,
1127
+ "model.layers.35.self_attn.o_proj": true,
1128
+ "model.layers.35.self_attn.rope": false,
1129
+ "model.layers.35.block_sparse_moe.gate": true,
1130
+ "model.layers.35.block_sparse_moe.switch_mlp.gate_proj": true,
1131
+ "model.layers.35.block_sparse_moe.switch_mlp.up_proj": true,
1132
+ "model.layers.35.block_sparse_moe.switch_mlp.down_proj": true,
1133
+ "model.layers.35.block_sparse_moe.switch_mlp.activation": false,
1134
+ "model.layers.35.input_layernorm": false,
1135
+ "model.layers.35.post_attention_layernorm": false,
1136
+ "model.layers.36.self_attn.q_proj": true,
1137
+ "model.layers.36.self_attn.k_proj": true,
1138
+ "model.layers.36.self_attn.v_proj": true,
1139
+ "model.layers.36.self_attn.o_proj": true,
1140
+ "model.layers.36.self_attn.rope": false,
1141
+ "model.layers.36.block_sparse_moe.gate": true,
1142
+ "model.layers.36.block_sparse_moe.switch_mlp.gate_proj": true,
1143
+ "model.layers.36.block_sparse_moe.switch_mlp.up_proj": true,
1144
+ "model.layers.36.block_sparse_moe.switch_mlp.down_proj": true,
1145
+ "model.layers.36.block_sparse_moe.switch_mlp.activation": false,
1146
+ "model.layers.36.input_layernorm": false,
1147
+ "model.layers.36.post_attention_layernorm": false,
1148
+ "model.layers.37.self_attn.q_proj": true,
1149
+ "model.layers.37.self_attn.k_proj": true,
1150
+ "model.layers.37.self_attn.v_proj": true,
1151
+ "model.layers.37.self_attn.o_proj": true,
1152
+ "model.layers.37.self_attn.rope": false,
1153
+ "model.layers.37.block_sparse_moe.gate": true,
1154
+ "model.layers.37.block_sparse_moe.switch_mlp.gate_proj": true,
1155
+ "model.layers.37.block_sparse_moe.switch_mlp.up_proj": true,
1156
+ "model.layers.37.block_sparse_moe.switch_mlp.down_proj": true,
1157
+ "model.layers.37.block_sparse_moe.switch_mlp.activation": false,
1158
+ "model.layers.37.input_layernorm": false,
1159
+ "model.layers.37.post_attention_layernorm": false,
1160
+ "model.layers.38.self_attn.q_proj": true,
1161
+ "model.layers.38.self_attn.k_proj": true,
1162
+ "model.layers.38.self_attn.v_proj": true,
1163
+ "model.layers.38.self_attn.o_proj": true,
1164
+ "model.layers.38.self_attn.rope": false,
1165
+ "model.layers.38.block_sparse_moe.gate": true,
1166
+ "model.layers.38.block_sparse_moe.switch_mlp.gate_proj": true,
1167
+ "model.layers.38.block_sparse_moe.switch_mlp.up_proj": true,
1168
+ "model.layers.38.block_sparse_moe.switch_mlp.down_proj": true,
1169
+ "model.layers.38.block_sparse_moe.switch_mlp.activation": false,
1170
+ "model.layers.38.input_layernorm": false,
1171
+ "model.layers.38.post_attention_layernorm": false,
1172
+ "model.layers.39.self_attn.q_proj": true,
1173
+ "model.layers.39.self_attn.k_proj": true,
1174
+ "model.layers.39.self_attn.v_proj": true,
1175
+ "model.layers.39.self_attn.o_proj": true,
1176
+ "model.layers.39.self_attn.rope": false,
1177
+ "model.layers.39.block_sparse_moe.gate": true,
1178
+ "model.layers.39.block_sparse_moe.switch_mlp.gate_proj": true,
1179
+ "model.layers.39.block_sparse_moe.switch_mlp.up_proj": true,
1180
+ "model.layers.39.block_sparse_moe.switch_mlp.down_proj": true,
1181
+ "model.layers.39.block_sparse_moe.switch_mlp.activation": false,
1182
+ "model.layers.39.input_layernorm": false,
1183
+ "model.layers.39.post_attention_layernorm": false,
1184
+ "model.layers.40.self_attn.q_proj": true,
1185
+ "model.layers.40.self_attn.k_proj": true,
1186
+ "model.layers.40.self_attn.v_proj": true,
1187
+ "model.layers.40.self_attn.o_proj": true,
1188
+ "model.layers.40.self_attn.rope": false,
1189
+ "model.layers.40.block_sparse_moe.gate": true,
1190
+ "model.layers.40.block_sparse_moe.switch_mlp.gate_proj": true,
1191
+ "model.layers.40.block_sparse_moe.switch_mlp.up_proj": true,
1192
+ "model.layers.40.block_sparse_moe.switch_mlp.down_proj": true,
1193
+ "model.layers.40.block_sparse_moe.switch_mlp.activation": false,
1194
+ "model.layers.40.input_layernorm": false,
1195
+ "model.layers.40.post_attention_layernorm": false,
1196
+ "model.layers.41.self_attn.q_proj": true,
1197
+ "model.layers.41.self_attn.k_proj": true,
1198
+ "model.layers.41.self_attn.v_proj": true,
1199
+ "model.layers.41.self_attn.o_proj": true,
1200
+ "model.layers.41.self_attn.rope": false,
1201
+ "model.layers.41.block_sparse_moe.gate": true,
1202
+ "model.layers.41.block_sparse_moe.switch_mlp.gate_proj": true,
1203
+ "model.layers.41.block_sparse_moe.switch_mlp.up_proj": true,
1204
+ "model.layers.41.block_sparse_moe.switch_mlp.down_proj": true,
1205
+ "model.layers.41.block_sparse_moe.switch_mlp.activation": false,
1206
+ "model.layers.41.input_layernorm": false,
1207
+ "model.layers.41.post_attention_layernorm": false,
1208
+ "model.layers.42.self_attn.q_proj": true,
1209
+ "model.layers.42.self_attn.k_proj": true,
1210
+ "model.layers.42.self_attn.v_proj": true,
1211
+ "model.layers.42.self_attn.o_proj": true,
1212
+ "model.layers.42.self_attn.rope": false,
1213
+ "model.layers.42.block_sparse_moe.gate": true,
1214
+ "model.layers.42.block_sparse_moe.switch_mlp.gate_proj": true,
1215
+ "model.layers.42.block_sparse_moe.switch_mlp.up_proj": true,
1216
+ "model.layers.42.block_sparse_moe.switch_mlp.down_proj": true,
1217
+ "model.layers.42.block_sparse_moe.switch_mlp.activation": false,
1218
+ "model.layers.42.input_layernorm": false,
1219
+ "model.layers.42.post_attention_layernorm": false,
1220
+ "model.layers.43.self_attn.q_proj": true,
1221
+ "model.layers.43.self_attn.k_proj": true,
1222
+ "model.layers.43.self_attn.v_proj": true,
1223
+ "model.layers.43.self_attn.o_proj": true,
1224
+ "model.layers.43.self_attn.rope": false,
1225
+ "model.layers.43.block_sparse_moe.gate": true,
1226
+ "model.layers.43.block_sparse_moe.switch_mlp.gate_proj": true,
1227
+ "model.layers.43.block_sparse_moe.switch_mlp.up_proj": true,
1228
+ "model.layers.43.block_sparse_moe.switch_mlp.down_proj": true,
1229
+ "model.layers.43.block_sparse_moe.switch_mlp.activation": false,
1230
+ "model.layers.43.input_layernorm": false,
1231
+ "model.layers.43.post_attention_layernorm": false,
1232
+ "model.layers.44.self_attn.q_proj": true,
1233
+ "model.layers.44.self_attn.k_proj": true,
1234
+ "model.layers.44.self_attn.v_proj": true,
1235
+ "model.layers.44.self_attn.o_proj": true,
1236
+ "model.layers.44.self_attn.rope": false,
1237
+ "model.layers.44.block_sparse_moe.gate": true,
1238
+ "model.layers.44.block_sparse_moe.switch_mlp.gate_proj": true,
1239
+ "model.layers.44.block_sparse_moe.switch_mlp.up_proj": true,
1240
+ "model.layers.44.block_sparse_moe.switch_mlp.down_proj": true,
1241
+ "model.layers.44.block_sparse_moe.switch_mlp.activation": false,
1242
+ "model.layers.44.input_layernorm": false,
1243
+ "model.layers.44.post_attention_layernorm": false,
1244
+ "model.layers.45.self_attn.q_proj": true,
1245
+ "model.layers.45.self_attn.k_proj": true,
1246
+ "model.layers.45.self_attn.v_proj": true,
1247
+ "model.layers.45.self_attn.o_proj": true,
1248
+ "model.layers.45.self_attn.rope": false,
1249
+ "model.layers.45.block_sparse_moe.gate": true,
1250
+ "model.layers.45.block_sparse_moe.switch_mlp.gate_proj": true,
1251
+ "model.layers.45.block_sparse_moe.switch_mlp.up_proj": true,
1252
+ "model.layers.45.block_sparse_moe.switch_mlp.down_proj": true,
1253
+ "model.layers.45.block_sparse_moe.switch_mlp.activation": false,
1254
+ "model.layers.45.input_layernorm": false,
1255
+ "model.layers.45.post_attention_layernorm": false,
1256
+ "model.layers.46.self_attn.q_proj": true,
1257
+ "model.layers.46.self_attn.k_proj": true,
1258
+ "model.layers.46.self_attn.v_proj": true,
1259
+ "model.layers.46.self_attn.o_proj": true,
1260
+ "model.layers.46.self_attn.rope": false,
1261
+ "model.layers.46.block_sparse_moe.gate": true,
1262
+ "model.layers.46.block_sparse_moe.switch_mlp.gate_proj": true,
1263
+ "model.layers.46.block_sparse_moe.switch_mlp.up_proj": true,
1264
+ "model.layers.46.block_sparse_moe.switch_mlp.down_proj": true,
1265
+ "model.layers.46.block_sparse_moe.switch_mlp.activation": false,
1266
+ "model.layers.46.input_layernorm": false,
1267
+ "model.layers.46.post_attention_layernorm": false,
1268
+ "model.layers.47.self_attn.q_proj": true,
1269
+ "model.layers.47.self_attn.k_proj": true,
1270
+ "model.layers.47.self_attn.v_proj": true,
1271
+ "model.layers.47.self_attn.o_proj": true,
1272
+ "model.layers.47.self_attn.rope": false,
1273
+ "model.layers.47.block_sparse_moe.gate": true,
1274
+ "model.layers.47.block_sparse_moe.switch_mlp.gate_proj": true,
1275
+ "model.layers.47.block_sparse_moe.switch_mlp.up_proj": true,
1276
+ "model.layers.47.block_sparse_moe.switch_mlp.down_proj": true,
1277
+ "model.layers.47.block_sparse_moe.switch_mlp.activation": false,
1278
+ "model.layers.47.input_layernorm": false,
1279
+ "model.layers.47.post_attention_layernorm": false,
1280
+ "model.layers.48.self_attn.q_proj": true,
1281
+ "model.layers.48.self_attn.k_proj": true,
1282
+ "model.layers.48.self_attn.v_proj": true,
1283
+ "model.layers.48.self_attn.o_proj": true,
1284
+ "model.layers.48.self_attn.rope": false,
1285
+ "model.layers.48.block_sparse_moe.gate": true,
1286
+ "model.layers.48.block_sparse_moe.switch_mlp.gate_proj": true,
1287
+ "model.layers.48.block_sparse_moe.switch_mlp.up_proj": true,
1288
+ "model.layers.48.block_sparse_moe.switch_mlp.down_proj": true,
1289
+ "model.layers.48.block_sparse_moe.switch_mlp.activation": false,
1290
+ "model.layers.48.input_layernorm": false,
1291
+ "model.layers.48.post_attention_layernorm": false,
1292
+ "model.layers.49.self_attn.q_proj": true,
1293
+ "model.layers.49.self_attn.k_proj": true,
1294
+ "model.layers.49.self_attn.v_proj": true,
1295
+ "model.layers.49.self_attn.o_proj": true,
1296
+ "model.layers.49.self_attn.rope": false,
1297
+ "model.layers.49.block_sparse_moe.gate": true,
1298
+ "model.layers.49.block_sparse_moe.switch_mlp.gate_proj": true,
1299
+ "model.layers.49.block_sparse_moe.switch_mlp.up_proj": true,
1300
+ "model.layers.49.block_sparse_moe.switch_mlp.down_proj": true,
1301
+ "model.layers.49.block_sparse_moe.switch_mlp.activation": false,
1302
+ "model.layers.49.input_layernorm": false,
1303
+ "model.layers.49.post_attention_layernorm": false,
1304
+ "model.layers.50.self_attn.q_proj": true,
1305
+ "model.layers.50.self_attn.k_proj": true,
1306
+ "model.layers.50.self_attn.v_proj": true,
1307
+ "model.layers.50.self_attn.o_proj": true,
1308
+ "model.layers.50.self_attn.rope": false,
1309
+ "model.layers.50.block_sparse_moe.gate": true,
1310
+ "model.layers.50.block_sparse_moe.switch_mlp.gate_proj": true,
1311
+ "model.layers.50.block_sparse_moe.switch_mlp.up_proj": true,
1312
+ "model.layers.50.block_sparse_moe.switch_mlp.down_proj": true,
1313
+ "model.layers.50.block_sparse_moe.switch_mlp.activation": false,
1314
+ "model.layers.50.input_layernorm": false,
1315
+ "model.layers.50.post_attention_layernorm": false,
1316
+ "model.layers.51.self_attn.q_proj": true,
1317
+ "model.layers.51.self_attn.k_proj": true,
1318
+ "model.layers.51.self_attn.v_proj": true,
1319
+ "model.layers.51.self_attn.o_proj": true,
1320
+ "model.layers.51.self_attn.rope": false,
1321
+ "model.layers.51.block_sparse_moe.gate": true,
1322
+ "model.layers.51.block_sparse_moe.switch_mlp.gate_proj": true,
1323
+ "model.layers.51.block_sparse_moe.switch_mlp.up_proj": true,
1324
+ "model.layers.51.block_sparse_moe.switch_mlp.down_proj": true,
1325
+ "model.layers.51.block_sparse_moe.switch_mlp.activation": false,
1326
+ "model.layers.51.input_layernorm": false,
1327
+ "model.layers.51.post_attention_layernorm": false,
1328
+ "model.layers.52.self_attn.q_proj": true,
1329
+ "model.layers.52.self_attn.k_proj": true,
1330
+ "model.layers.52.self_attn.v_proj": true,
1331
+ "model.layers.52.self_attn.o_proj": true,
1332
+ "model.layers.52.self_attn.rope": false,
1333
+ "model.layers.52.block_sparse_moe.gate": true,
1334
+ "model.layers.52.block_sparse_moe.switch_mlp.gate_proj": true,
1335
+ "model.layers.52.block_sparse_moe.switch_mlp.up_proj": true,
1336
+ "model.layers.52.block_sparse_moe.switch_mlp.down_proj": true,
1337
+ "model.layers.52.block_sparse_moe.switch_mlp.activation": false,
1338
+ "model.layers.52.input_layernorm": false,
1339
+ "model.layers.52.post_attention_layernorm": false,
1340
+ "model.layers.53.self_attn.q_proj": true,
1341
+ "model.layers.53.self_attn.k_proj": true,
1342
+ "model.layers.53.self_attn.v_proj": true,
1343
+ "model.layers.53.self_attn.o_proj": true,
1344
+ "model.layers.53.self_attn.rope": false,
1345
+ "model.layers.53.block_sparse_moe.gate": true,
1346
+ "model.layers.53.block_sparse_moe.switch_mlp.gate_proj": true,
1347
+ "model.layers.53.block_sparse_moe.switch_mlp.up_proj": true,
1348
+ "model.layers.53.block_sparse_moe.switch_mlp.down_proj": true,
1349
+ "model.layers.53.block_sparse_moe.switch_mlp.activation": false,
1350
+ "model.layers.53.input_layernorm": false,
1351
+ "model.layers.53.post_attention_layernorm": false,
1352
+ "model.layers.54.self_attn.q_proj": true,
1353
+ "model.layers.54.self_attn.k_proj": true,
1354
+ "model.layers.54.self_attn.v_proj": true,
1355
+ "model.layers.54.self_attn.o_proj": true,
1356
+ "model.layers.54.self_attn.rope": false,
1357
+ "model.layers.54.block_sparse_moe.gate": true,
1358
+ "model.layers.54.block_sparse_moe.switch_mlp.gate_proj": true,
1359
+ "model.layers.54.block_sparse_moe.switch_mlp.up_proj": true,
1360
+ "model.layers.54.block_sparse_moe.switch_mlp.down_proj": true,
1361
+ "model.layers.54.block_sparse_moe.switch_mlp.activation": false,
1362
+ "model.layers.54.input_layernorm": false,
1363
+ "model.layers.54.post_attention_layernorm": false,
1364
+ "model.layers.55.self_attn.q_proj": true,
1365
+ "model.layers.55.self_attn.k_proj": true,
1366
+ "model.layers.55.self_attn.v_proj": true,
1367
+ "model.layers.55.self_attn.o_proj": true,
1368
+ "model.layers.55.self_attn.rope": false,
1369
+ "model.layers.55.block_sparse_moe.gate": true,
1370
+ "model.layers.55.block_sparse_moe.switch_mlp.gate_proj": true,
1371
+ "model.layers.55.block_sparse_moe.switch_mlp.up_proj": true,
1372
+ "model.layers.55.block_sparse_moe.switch_mlp.down_proj": true,
1373
+ "model.layers.55.block_sparse_moe.switch_mlp.activation": false,
1374
+ "model.layers.55.input_layernorm": false,
1375
+ "model.layers.55.post_attention_layernorm": false,
1376
+ "model.norm": false,
1377
+ "lm_head": true
1378
+ },
1379
+ "rms_norm_eps": 1e-05,
1380
+ "rope_theta": 1000000.0,
1381
+ "router_aux_loss_coef": 0.001,
1382
+ "router_jitter_noise": 0.0,
1383
+ "sliding_window": null,
1384
+ "tie_word_embeddings": false,
1385
+ "torch_dtype": "bfloat16",
1386
+ "transformers_version": "4.52.0.dev0",
1387
+ "use_cache": true,
1388
+ "vocab_size": 32768
1389
+ }
model-00001-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8daf04e6e7dcdbfc265181569e57ed6e2cd651ecc3aec14d919cd0aef4651ce
3
+ size 5337187942
model-00002-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f712a8c94b3dece543396e1aaf59b0317ce02e0a40b3b98de7e51d86eaf8f99c
3
+ size 5337213006
model-00003-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:86ae270c39d19d8cb7f9128fe8af06e06c497684397dd264d98e35afc7ef9c50
3
+ size 5287634399
model-00004-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de27eabadf9ffc738a3b5d8827e538e6e73b44b2ef2d56da9692d78fbf53ff30
3
+ size 5337213074
model-00005-of-00005.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5d4c34574286fb388d2bac4cdb75215a9da6c3d107f9e8ff8c4cde1b72907b3
3
+ size 729095424
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<s>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9addc8bdce5988448ae81b729336f43a81262160ae8da760674badab9d4c7d33
3
+ size 587591
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff