Hugging Face's logo

bisonnetworking
/
MediPhi-Instruct-mlx-4bit

Text Generation
MLX
Safetensors
English
phi3
mlx-lm
phi
phi-3
medical
clinical
healthcare
quantized
4bit
on-device
ios
apple-silicon
conversational
4-bit precision
Model card Files
xet
Community
bisonnetworking commited on
Commit
6b84460
·
verified ·
1 Parent(s): a7da84b

Upload folder using huggingface_hub

Browse files
Files changed (8) hide show
  1. README.md +77 -0
  2. chat_template.jinja +8 -0
  3. config.json +143 -0
  4. generation_config.json +7 -0
  5. model.safetensors +3 -0
  6. model.safetensors.index.json +463 -0
  7. tokenizer.json +0 -0
  8. tokenizer_config.json +16 -0
README.md ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ base_model: microsoft/MediPhi-Instruct
4
+ library_name: mlx
5
+ pipeline_tag: text-generation
6
+ language:
7
+ - en
8
+
9
+ tags:
10
+ - mlx
11
+ - mlx-lm
12
+ - phi
13
+ - phi-3
14
+ - medical
15
+ - clinical
16
+ - healthcare
17
+ - quantized
18
+ - 4bit
19
+ - on-device
20
+ - ios
21
+ - apple-silicon
22
+
23
+ model_type: phi
24
+ quantization:
25
+ - 4bit
26
+
27
+ datasets:
28
+ - microsoft/mediflow
29
+ - ncbi/pubmed
30
+ - epfl-llm/guidelines
31
+ - starmpcc/Asclepius-Synthetic-Clinical-Notes
32
+ - akemiH/NoteChat
33
+ - zhengyun21/PMC-Patients
34
+ - jpcorb20/medical_wikipedia
35
+ ---
36
+
37
+ # MediPhi-Instruct (MLX · 4-bit)
38
+
39
+ This repository contains an **MLX-format 4-bit quantized** version of
40
+ [`microsoft/MediPhi-Instruct`](https://huggingface.co/microsoft/MediPhi-Instruct),
41
+ converted using **`mlx-lm`** for efficient **on-device inference** on Apple silicon.
42
+
43
+ This model is intended for **iOS / iPadOS / macOS** usage where memory and power
44
+ constraints require aggressive quantization while preserving clinical reasoning quality.
45
+
46
+ ---
47
+
48
+ ## Model details
49
+
50
+ - **Base model:** MediPhi-Instruct (Phi-3 family)
51
+ - **Parameters:** ~3.8B
52
+ - **Quantization:** 4-bit (MLX)
53
+ - **Format:** MLX (not GGUF)
54
+ - **Intended use:** On-device medical and clinical QA, decision support, and explanations
55
+ - **Language:** English
56
+
57
+ > ⚠️ This is a **conversion only**. No additional fine-tuning was performed.
58
+
59
+ ---
60
+
61
+ ## Why MLX 4-bit?
62
+
63
+ Compared to larger 4–7B medical models, MediPhi-Instruct shows:
64
+ - Strong clinical reasoning per parameter
65
+ - Better robustness under 4-bit quantization
66
+ - Lower memory footprint suitable for mobile devices
67
+
68
+ This makes it a strong candidate for **on-device medical assistants** on iPhone and iPad.
69
+
70
+ ---
71
+
72
+ ## Usage (MLX-LM)
73
+
74
+ ### Install
75
+ ```bash
76
+ pip install mlx-lm
77
+
chat_template.jinja ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {% for message in messages %}{% if message['role'] == 'system' and message['content'] %}{{'<|system|>
2
+ ' + message['content'] + '<|end|>
3
+ '}}{% elif message['role'] == 'user' %}{{'<|user|>
4
+ ' + message['content'] + '<|end|>
5
+ '}}{% elif message['role'] == 'assistant' %}{{'<|assistant|>
6
+ ' + message['content'] + '<|end|>
7
+ '}}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '<|assistant|>
8
+ ' }}{% else %}{{ eos_token }}{% endif %}
config.json ADDED
@@ -0,0 +1,143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Phi3ForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "embd_pdrop": 0.0,
9
+ "eos_token_id": 32000,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 3072,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 8192,
14
+ "max_position_embeddings": 131072,
15
+ "model_type": "phi3",
16
+ "num_attention_heads": 32,
17
+ "num_hidden_layers": 32,
18
+ "num_key_value_heads": 32,
19
+ "original_max_position_embeddings": 4096,
20
+ "pad_token_id": 32000,
21
+ "quantization": {
22
+ "group_size": 64,
23
+ "bits": 4,
24
+ "mode": "affine"
25
+ },
26
+ "quantization_config": {
27
+ "group_size": 64,
28
+ "bits": 4,
29
+ "mode": "affine"
30
+ },
31
+ "resid_pdrop": 0.0,
32
+ "rms_norm_eps": 1e-05,
33
+ "rope_scaling": {
34
+ "long_factor": [
35
+ 1.0800000429153442,
36
+ 1.1100000143051147,
37
+ 1.1399999856948853,
38
+ 1.340000033378601,
39
+ 1.5899999141693115,
40
+ 1.600000023841858,
41
+ 1.6200000047683716,
42
+ 2.620000123977661,
43
+ 3.2300000190734863,
44
+ 3.2300000190734863,
45
+ 4.789999961853027,
46
+ 7.400000095367432,
47
+ 7.700000286102295,
48
+ 9.09000015258789,
49
+ 12.199999809265137,
50
+ 17.670000076293945,
51
+ 24.46000099182129,
52
+ 28.57000160217285,
53
+ 30.420001983642578,
54
+ 30.840002059936523,
55
+ 32.590003967285156,
56
+ 32.93000411987305,
57
+ 42.320003509521484,
58
+ 44.96000289916992,
59
+ 50.340003967285156,
60
+ 50.45000457763672,
61
+ 57.55000305175781,
62
+ 57.93000411987305,
63
+ 58.21000289916992,
64
+ 60.1400032043457,
65
+ 62.61000442504883,
66
+ 62.62000274658203,
67
+ 62.71000289916992,
68
+ 63.1400032043457,
69
+ 63.1400032043457,
70
+ 63.77000427246094,
71
+ 63.93000411987305,
72
+ 63.96000289916992,
73
+ 63.970001220703125,
74
+ 64.02999877929688,
75
+ 64.06999969482422,
76
+ 64.08000183105469,
77
+ 64.12000274658203,
78
+ 64.41000366210938,
79
+ 64.4800033569336,
80
+ 64.51000213623047,
81
+ 64.52999877929688,
82
+ 64.83999633789062
83
+ ],
84
+ "short_factor": [
85
+ 1.0,
86
+ 1.0199999809265137,
87
+ 1.0299999713897705,
88
+ 1.0299999713897705,
89
+ 1.0499999523162842,
90
+ 1.0499999523162842,
91
+ 1.0499999523162842,
92
+ 1.0499999523162842,
93
+ 1.0499999523162842,
94
+ 1.0699999332427979,
95
+ 1.0999999046325684,
96
+ 1.1099998950958252,
97
+ 1.1599998474121094,
98
+ 1.1599998474121094,
99
+ 1.1699998378753662,
100
+ 1.2899998426437378,
101
+ 1.339999794960022,
102
+ 1.679999828338623,
103
+ 1.7899998426437378,
104
+ 1.8199998140335083,
105
+ 1.8499997854232788,
106
+ 1.8799997568130493,
107
+ 1.9099997282028198,
108
+ 1.9399996995925903,
109
+ 1.9899996519088745,
110
+ 2.0199997425079346,
111
+ 2.0199997425079346,
112
+ 2.0199997425079346,
113
+ 2.0199997425079346,
114
+ 2.0199997425079346,
115
+ 2.0199997425079346,
116
+ 2.0299997329711914,
117
+ 2.0299997329711914,
118
+ 2.0299997329711914,
119
+ 2.0299997329711914,
120
+ 2.0299997329711914,
121
+ 2.0299997329711914,
122
+ 2.0299997329711914,
123
+ 2.0299997329711914,
124
+ 2.0299997329711914,
125
+ 2.0799996852874756,
126
+ 2.0899996757507324,
127
+ 2.189999580383301,
128
+ 2.2199995517730713,
129
+ 2.5899994373321533,
130
+ 2.729999542236328,
131
+ 2.749999523162842,
132
+ 2.8399994373321533
133
+ ],
134
+ "type": "longrope"
135
+ },
136
+ "rope_theta": 10000.0,
137
+ "sliding_window": 262144,
138
+ "tie_word_embeddings": false,
139
+ "torch_dtype": "bfloat16",
140
+ "transformers_version": "4.47.0",
141
+ "use_cache": false,
142
+ "vocab_size": 32064
143
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 32000,
5
+ "pad_token_id": 32000,
6
+ "transformers_version": "4.47.0"
7
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ead301b4546478fb4700138a12ebe98ce686f554237990e4be8919d36d34397
3
+ size 2149696492
model.safetensors.index.json ADDED
@@ -0,0 +1,463 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 2149644288,
4
+ "total_parameters": 3821079552
5
+ },
6
+ "weight_map": {
7
+ "lm_head.biases": "model.safetensors",
8
+ "lm_head.scales": "model.safetensors",
9
+ "lm_head.weight": "model.safetensors",
10
+ "model.embed_tokens.biases": "model.safetensors",
11
+ "model.embed_tokens.scales": "model.safetensors",
12
+ "model.embed_tokens.weight": "model.safetensors",
13
+ "model.layers.0.input_layernorm.weight": "model.safetensors",
14
+ "model.layers.0.mlp.down_proj.biases": "model.safetensors",
15
+ "model.layers.0.mlp.down_proj.scales": "model.safetensors",
16
+ "model.layers.0.mlp.down_proj.weight": "model.safetensors",
17
+ "model.layers.0.mlp.gate_up_proj.biases": "model.safetensors",
18
+ "model.layers.0.mlp.gate_up_proj.scales": "model.safetensors",
19
+ "model.layers.0.mlp.gate_up_proj.weight": "model.safetensors",
20
+ "model.layers.0.post_attention_layernorm.weight": "model.safetensors",
21
+ "model.layers.0.self_attn.o_proj.biases": "model.safetensors",
22
+ "model.layers.0.self_attn.o_proj.scales": "model.safetensors",
23
+ "model.layers.0.self_attn.o_proj.weight": "model.safetensors",
24
+ "model.layers.0.self_attn.qkv_proj.biases": "model.safetensors",
25
+ "model.layers.0.self_attn.qkv_proj.scales": "model.safetensors",
26
+ "model.layers.0.self_attn.qkv_proj.weight": "model.safetensors",
27
+ "model.layers.1.input_layernorm.weight": "model.safetensors",
28
+ "model.layers.1.mlp.down_proj.biases": "model.safetensors",
29
+ "model.layers.1.mlp.down_proj.scales": "model.safetensors",
30
+ "model.layers.1.mlp.down_proj.weight": "model.safetensors",
31
+ "model.layers.1.mlp.gate_up_proj.biases": "model.safetensors",
32
+ "model.layers.1.mlp.gate_up_proj.scales": "model.safetensors",
33
+ "model.layers.1.mlp.gate_up_proj.weight": "model.safetensors",
34
+ "model.layers.1.post_attention_layernorm.weight": "model.safetensors",
35
+ "model.layers.1.self_attn.o_proj.biases": "model.safetensors",
36
+ "model.layers.1.self_attn.o_proj.scales": "model.safetensors",
37
+ "model.layers.1.self_attn.o_proj.weight": "model.safetensors",
38
+ "model.layers.1.self_attn.qkv_proj.biases": "model.safetensors",
39
+ "model.layers.1.self_attn.qkv_proj.scales": "model.safetensors",
40
+ "model.layers.1.self_attn.qkv_proj.weight": "model.safetensors",
41
+ "model.layers.10.input_layernorm.weight": "model.safetensors",
42
+ "model.layers.10.mlp.down_proj.biases": "model.safetensors",
43
+ "model.layers.10.mlp.down_proj.scales": "model.safetensors",
44
+ "model.layers.10.mlp.down_proj.weight": "model.safetensors",
45
+ "model.layers.10.mlp.gate_up_proj.biases": "model.safetensors",
46
+ "model.layers.10.mlp.gate_up_proj.scales": "model.safetensors",
47
+ "model.layers.10.mlp.gate_up_proj.weight": "model.safetensors",
48
+ "model.layers.10.post_attention_layernorm.weight": "model.safetensors",
49
+ "model.layers.10.self_attn.o_proj.biases": "model.safetensors",
50
+ "model.layers.10.self_attn.o_proj.scales": "model.safetensors",
51
+ "model.layers.10.self_attn.o_proj.weight": "model.safetensors",
52
+ "model.layers.10.self_attn.qkv_proj.biases": "model.safetensors",
53
+ "model.layers.10.self_attn.qkv_proj.scales": "model.safetensors",
54
+ "model.layers.10.self_attn.qkv_proj.weight": "model.safetensors",
55
+ "model.layers.11.input_layernorm.weight": "model.safetensors",
56
+ "model.layers.11.mlp.down_proj.biases": "model.safetensors",
57
+ "model.layers.11.mlp.down_proj.scales": "model.safetensors",
58
+ "model.layers.11.mlp.down_proj.weight": "model.safetensors",
59
+ "model.layers.11.mlp.gate_up_proj.biases": "model.safetensors",
60
+ "model.layers.11.mlp.gate_up_proj.scales": "model.safetensors",
61
+ "model.layers.11.mlp.gate_up_proj.weight": "model.safetensors",
62
+ "model.layers.11.post_attention_layernorm.weight": "model.safetensors",
63
+ "model.layers.11.self_attn.o_proj.biases": "model.safetensors",
64
+ "model.layers.11.self_attn.o_proj.scales": "model.safetensors",
65
+ "model.layers.11.self_attn.o_proj.weight": "model.safetensors",
66
+ "model.layers.11.self_attn.qkv_proj.biases": "model.safetensors",
67
+ "model.layers.11.self_attn.qkv_proj.scales": "model.safetensors",
68
+ "model.layers.11.self_attn.qkv_proj.weight": "model.safetensors",
69
+ "model.layers.12.input_layernorm.weight": "model.safetensors",
70
+ "model.layers.12.mlp.down_proj.biases": "model.safetensors",
71
+ "model.layers.12.mlp.down_proj.scales": "model.safetensors",
72
+ "model.layers.12.mlp.down_proj.weight": "model.safetensors",
73
+ "model.layers.12.mlp.gate_up_proj.biases": "model.safetensors",
74
+ "model.layers.12.mlp.gate_up_proj.scales": "model.safetensors",
75
+ "model.layers.12.mlp.gate_up_proj.weight": "model.safetensors",
76
+ "model.layers.12.post_attention_layernorm.weight": "model.safetensors",
77
+ "model.layers.12.self_attn.o_proj.biases": "model.safetensors",
78
+ "model.layers.12.self_attn.o_proj.scales": "model.safetensors",
79
+ "model.layers.12.self_attn.o_proj.weight": "model.safetensors",
80
+ "model.layers.12.self_attn.qkv_proj.biases": "model.safetensors",
81
+ "model.layers.12.self_attn.qkv_proj.scales": "model.safetensors",
82
+ "model.layers.12.self_attn.qkv_proj.weight": "model.safetensors",
83
+ "model.layers.13.input_layernorm.weight": "model.safetensors",
84
+ "model.layers.13.mlp.down_proj.biases": "model.safetensors",
85
+ "model.layers.13.mlp.down_proj.scales": "model.safetensors",
86
+ "model.layers.13.mlp.down_proj.weight": "model.safetensors",
87
+ "model.layers.13.mlp.gate_up_proj.biases": "model.safetensors",
88
+ "model.layers.13.mlp.gate_up_proj.scales": "model.safetensors",
89
+ "model.layers.13.mlp.gate_up_proj.weight": "model.safetensors",
90
+ "model.layers.13.post_attention_layernorm.weight": "model.safetensors",
91
+ "model.layers.13.self_attn.o_proj.biases": "model.safetensors",
92
+ "model.layers.13.self_attn.o_proj.scales": "model.safetensors",
93
+ "model.layers.13.self_attn.o_proj.weight": "model.safetensors",
94
+ "model.layers.13.self_attn.qkv_proj.biases": "model.safetensors",
95
+ "model.layers.13.self_attn.qkv_proj.scales": "model.safetensors",
96
+ "model.layers.13.self_attn.qkv_proj.weight": "model.safetensors",
97
+ "model.layers.14.input_layernorm.weight": "model.safetensors",
98
+ "model.layers.14.mlp.down_proj.biases": "model.safetensors",
99
+ "model.layers.14.mlp.down_proj.scales": "model.safetensors",
100
+ "model.layers.14.mlp.down_proj.weight": "model.safetensors",
101
+ "model.layers.14.mlp.gate_up_proj.biases": "model.safetensors",
102
+ "model.layers.14.mlp.gate_up_proj.scales": "model.safetensors",
103
+ "model.layers.14.mlp.gate_up_proj.weight": "model.safetensors",
104
+ "model.layers.14.post_attention_layernorm.weight": "model.safetensors",
105
+ "model.layers.14.self_attn.o_proj.biases": "model.safetensors",
106
+ "model.layers.14.self_attn.o_proj.scales": "model.safetensors",
107
+ "model.layers.14.self_attn.o_proj.weight": "model.safetensors",
108
+ "model.layers.14.self_attn.qkv_proj.biases": "model.safetensors",
109
+ "model.layers.14.self_attn.qkv_proj.scales": "model.safetensors",
110
+ "model.layers.14.self_attn.qkv_proj.weight": "model.safetensors",
111
+ "model.layers.15.input_layernorm.weight": "model.safetensors",
112
+ "model.layers.15.mlp.down_proj.biases": "model.safetensors",
113
+ "model.layers.15.mlp.down_proj.scales": "model.safetensors",
114
+ "model.layers.15.mlp.down_proj.weight": "model.safetensors",
115
+ "model.layers.15.mlp.gate_up_proj.biases": "model.safetensors",
116
+ "model.layers.15.mlp.gate_up_proj.scales": "model.safetensors",
117
+ "model.layers.15.mlp.gate_up_proj.weight": "model.safetensors",
118
+ "model.layers.15.post_attention_layernorm.weight": "model.safetensors",
119
+ "model.layers.15.self_attn.o_proj.biases": "model.safetensors",
120
+ "model.layers.15.self_attn.o_proj.scales": "model.safetensors",
121
+ "model.layers.15.self_attn.o_proj.weight": "model.safetensors",
122
+ "model.layers.15.self_attn.qkv_proj.biases": "model.safetensors",
123
+ "model.layers.15.self_attn.qkv_proj.scales": "model.safetensors",
124
+ "model.layers.15.self_attn.qkv_proj.weight": "model.safetensors",
125
+ "model.layers.16.input_layernorm.weight": "model.safetensors",
126
+ "model.layers.16.mlp.down_proj.biases": "model.safetensors",
127
+ "model.layers.16.mlp.down_proj.scales": "model.safetensors",
128
+ "model.layers.16.mlp.down_proj.weight": "model.safetensors",
129
+ "model.layers.16.mlp.gate_up_proj.biases": "model.safetensors",
130
+ "model.layers.16.mlp.gate_up_proj.scales": "model.safetensors",
131
+ "model.layers.16.mlp.gate_up_proj.weight": "model.safetensors",
132
+ "model.layers.16.post_attention_layernorm.weight": "model.safetensors",
133
+ "model.layers.16.self_attn.o_proj.biases": "model.safetensors",
134
+ "model.layers.16.self_attn.o_proj.scales": "model.safetensors",
135
+ "model.layers.16.self_attn.o_proj.weight": "model.safetensors",
136
+ "model.layers.16.self_attn.qkv_proj.biases": "model.safetensors",
137
+ "model.layers.16.self_attn.qkv_proj.scales": "model.safetensors",
138
+ "model.layers.16.self_attn.qkv_proj.weight": "model.safetensors",
139
+ "model.layers.17.input_layernorm.weight": "model.safetensors",
140
+ "model.layers.17.mlp.down_proj.biases": "model.safetensors",
141
+ "model.layers.17.mlp.down_proj.scales": "model.safetensors",
142
+ "model.layers.17.mlp.down_proj.weight": "model.safetensors",
143
+ "model.layers.17.mlp.gate_up_proj.biases": "model.safetensors",
144
+ "model.layers.17.mlp.gate_up_proj.scales": "model.safetensors",
145
+ "model.layers.17.mlp.gate_up_proj.weight": "model.safetensors",
146
+ "model.layers.17.post_attention_layernorm.weight": "model.safetensors",
147
+ "model.layers.17.self_attn.o_proj.biases": "model.safetensors",
148
+ "model.layers.17.self_attn.o_proj.scales": "model.safetensors",
149
+ "model.layers.17.self_attn.o_proj.weight": "model.safetensors",
150
+ "model.layers.17.self_attn.qkv_proj.biases": "model.safetensors",
151
+ "model.layers.17.self_attn.qkv_proj.scales": "model.safetensors",
152
+ "model.layers.17.self_attn.qkv_proj.weight": "model.safetensors",
153
+ "model.layers.18.input_layernorm.weight": "model.safetensors",
154
+ "model.layers.18.mlp.down_proj.biases": "model.safetensors",
155
+ "model.layers.18.mlp.down_proj.scales": "model.safetensors",
156
+ "model.layers.18.mlp.down_proj.weight": "model.safetensors",
157
+ "model.layers.18.mlp.gate_up_proj.biases": "model.safetensors",
158
+ "model.layers.18.mlp.gate_up_proj.scales": "model.safetensors",
159
+ "model.layers.18.mlp.gate_up_proj.weight": "model.safetensors",
160
+ "model.layers.18.post_attention_layernorm.weight": "model.safetensors",
161
+ "model.layers.18.self_attn.o_proj.biases": "model.safetensors",
162
+ "model.layers.18.self_attn.o_proj.scales": "model.safetensors",
163
+ "model.layers.18.self_attn.o_proj.weight": "model.safetensors",
164
+ "model.layers.18.self_attn.qkv_proj.biases": "model.safetensors",
165
+ "model.layers.18.self_attn.qkv_proj.scales": "model.safetensors",
166
+ "model.layers.18.self_attn.qkv_proj.weight": "model.safetensors",
167
+ "model.layers.19.input_layernorm.weight": "model.safetensors",
168
+ "model.layers.19.mlp.down_proj.biases": "model.safetensors",
169
+ "model.layers.19.mlp.down_proj.scales": "model.safetensors",
170
+ "model.layers.19.mlp.down_proj.weight": "model.safetensors",
171
+ "model.layers.19.mlp.gate_up_proj.biases": "model.safetensors",
172
+ "model.layers.19.mlp.gate_up_proj.scales": "model.safetensors",
173
+ "model.layers.19.mlp.gate_up_proj.weight": "model.safetensors",
174
+ "model.layers.19.post_attention_layernorm.weight": "model.safetensors",
175
+ "model.layers.19.self_attn.o_proj.biases": "model.safetensors",
176
+ "model.layers.19.self_attn.o_proj.scales": "model.safetensors",
177
+ "model.layers.19.self_attn.o_proj.weight": "model.safetensors",
178
+ "model.layers.19.self_attn.qkv_proj.biases": "model.safetensors",
179
+ "model.layers.19.self_attn.qkv_proj.scales": "model.safetensors",
180
+ "model.layers.19.self_attn.qkv_proj.weight": "model.safetensors",
181
+ "model.layers.2.input_layernorm.weight": "model.safetensors",
182
+ "model.layers.2.mlp.down_proj.biases": "model.safetensors",
183
+ "model.layers.2.mlp.down_proj.scales": "model.safetensors",
184
+ "model.layers.2.mlp.down_proj.weight": "model.safetensors",
185
+ "model.layers.2.mlp.gate_up_proj.biases": "model.safetensors",
186
+ "model.layers.2.mlp.gate_up_proj.scales": "model.safetensors",
187
+ "model.layers.2.mlp.gate_up_proj.weight": "model.safetensors",
188
+ "model.layers.2.post_attention_layernorm.weight": "model.safetensors",
189
+ "model.layers.2.self_attn.o_proj.biases": "model.safetensors",
190
+ "model.layers.2.self_attn.o_proj.scales": "model.safetensors",
191
+ "model.layers.2.self_attn.o_proj.weight": "model.safetensors",
192
+ "model.layers.2.self_attn.qkv_proj.biases": "model.safetensors",
193
+ "model.layers.2.self_attn.qkv_proj.scales": "model.safetensors",
194
+ "model.layers.2.self_attn.qkv_proj.weight": "model.safetensors",
195
+ "model.layers.20.input_layernorm.weight": "model.safetensors",
196
+ "model.layers.20.mlp.down_proj.biases": "model.safetensors",
197
+ "model.layers.20.mlp.down_proj.scales": "model.safetensors",
198
+ "model.layers.20.mlp.down_proj.weight": "model.safetensors",
199
+ "model.layers.20.mlp.gate_up_proj.biases": "model.safetensors",
200
+ "model.layers.20.mlp.gate_up_proj.scales": "model.safetensors",
201
+ "model.layers.20.mlp.gate_up_proj.weight": "model.safetensors",
202
+ "model.layers.20.post_attention_layernorm.weight": "model.safetensors",
203
+ "model.layers.20.self_attn.o_proj.biases": "model.safetensors",
204
+ "model.layers.20.self_attn.o_proj.scales": "model.safetensors",
205
+ "model.layers.20.self_attn.o_proj.weight": "model.safetensors",
206
+ "model.layers.20.self_attn.qkv_proj.biases": "model.safetensors",
207
+ "model.layers.20.self_attn.qkv_proj.scales": "model.safetensors",
208
+ "model.layers.20.self_attn.qkv_proj.weight": "model.safetensors",
209
+ "model.layers.21.input_layernorm.weight": "model.safetensors",
210
+ "model.layers.21.mlp.down_proj.biases": "model.safetensors",
211
+ "model.layers.21.mlp.down_proj.scales": "model.safetensors",
212
+ "model.layers.21.mlp.down_proj.weight": "model.safetensors",
213
+ "model.layers.21.mlp.gate_up_proj.biases": "model.safetensors",
214
+ "model.layers.21.mlp.gate_up_proj.scales": "model.safetensors",
215
+ "model.layers.21.mlp.gate_up_proj.weight": "model.safetensors",
216
+ "model.layers.21.post_attention_layernorm.weight": "model.safetensors",
217
+ "model.layers.21.self_attn.o_proj.biases": "model.safetensors",
218
+ "model.layers.21.self_attn.o_proj.scales": "model.safetensors",
219
+ "model.layers.21.self_attn.o_proj.weight": "model.safetensors",
220
+ "model.layers.21.self_attn.qkv_proj.biases": "model.safetensors",
221
+ "model.layers.21.self_attn.qkv_proj.scales": "model.safetensors",
222
+ "model.layers.21.self_attn.qkv_proj.weight": "model.safetensors",
223
+ "model.layers.22.input_layernorm.weight": "model.safetensors",
224
+ "model.layers.22.mlp.down_proj.biases": "model.safetensors",
225
+ "model.layers.22.mlp.down_proj.scales": "model.safetensors",
226
+ "model.layers.22.mlp.down_proj.weight": "model.safetensors",
227
+ "model.layers.22.mlp.gate_up_proj.biases": "model.safetensors",
228
+ "model.layers.22.mlp.gate_up_proj.scales": "model.safetensors",
229
+ "model.layers.22.mlp.gate_up_proj.weight": "model.safetensors",
230
+ "model.layers.22.post_attention_layernorm.weight": "model.safetensors",
231
+ "model.layers.22.self_attn.o_proj.biases": "model.safetensors",
232
+ "model.layers.22.self_attn.o_proj.scales": "model.safetensors",
233
+ "model.layers.22.self_attn.o_proj.weight": "model.safetensors",
234
+ "model.layers.22.self_attn.qkv_proj.biases": "model.safetensors",
235
+ "model.layers.22.self_attn.qkv_proj.scales": "model.safetensors",
236
+ "model.layers.22.self_attn.qkv_proj.weight": "model.safetensors",
237
+ "model.layers.23.input_layernorm.weight": "model.safetensors",
238
+ "model.layers.23.mlp.down_proj.biases": "model.safetensors",
239
+ "model.layers.23.mlp.down_proj.scales": "model.safetensors",
240
+ "model.layers.23.mlp.down_proj.weight": "model.safetensors",
241
+ "model.layers.23.mlp.gate_up_proj.biases": "model.safetensors",
242
+ "model.layers.23.mlp.gate_up_proj.scales": "model.safetensors",
243
+ "model.layers.23.mlp.gate_up_proj.weight": "model.safetensors",
244
+ "model.layers.23.post_attention_layernorm.weight": "model.safetensors",
245
+ "model.layers.23.self_attn.o_proj.biases": "model.safetensors",
246
+ "model.layers.23.self_attn.o_proj.scales": "model.safetensors",
247
+ "model.layers.23.self_attn.o_proj.weight": "model.safetensors",
248
+ "model.layers.23.self_attn.qkv_proj.biases": "model.safetensors",
249
+ "model.layers.23.self_attn.qkv_proj.scales": "model.safetensors",
250
+ "model.layers.23.self_attn.qkv_proj.weight": "model.safetensors",
251
+ "model.layers.24.input_layernorm.weight": "model.safetensors",
252
+ "model.layers.24.mlp.down_proj.biases": "model.safetensors",
253
+ "model.layers.24.mlp.down_proj.scales": "model.safetensors",
254
+ "model.layers.24.mlp.down_proj.weight": "model.safetensors",
255
+ "model.layers.24.mlp.gate_up_proj.biases": "model.safetensors",
256
+ "model.layers.24.mlp.gate_up_proj.scales": "model.safetensors",
257
+ "model.layers.24.mlp.gate_up_proj.weight": "model.safetensors",
258
+ "model.layers.24.post_attention_layernorm.weight": "model.safetensors",
259
+ "model.layers.24.self_attn.o_proj.biases": "model.safetensors",
260
+ "model.layers.24.self_attn.o_proj.scales": "model.safetensors",
261
+ "model.layers.24.self_attn.o_proj.weight": "model.safetensors",
262
+ "model.layers.24.self_attn.qkv_proj.biases": "model.safetensors",
263
+ "model.layers.24.self_attn.qkv_proj.scales": "model.safetensors",
264
+ "model.layers.24.self_attn.qkv_proj.weight": "model.safetensors",
265
+ "model.layers.25.input_layernorm.weight": "model.safetensors",
266
+ "model.layers.25.mlp.down_proj.biases": "model.safetensors",
267
+ "model.layers.25.mlp.down_proj.scales": "model.safetensors",
268
+ "model.layers.25.mlp.down_proj.weight": "model.safetensors",
269
+ "model.layers.25.mlp.gate_up_proj.biases": "model.safetensors",
270
+ "model.layers.25.mlp.gate_up_proj.scales": "model.safetensors",
271
+ "model.layers.25.mlp.gate_up_proj.weight": "model.safetensors",
272
+ "model.layers.25.post_attention_layernorm.weight": "model.safetensors",
273
+ "model.layers.25.self_attn.o_proj.biases": "model.safetensors",
274
+ "model.layers.25.self_attn.o_proj.scales": "model.safetensors",
275
+ "model.layers.25.self_attn.o_proj.weight": "model.safetensors",
276
+ "model.layers.25.self_attn.qkv_proj.biases": "model.safetensors",
277
+ "model.layers.25.self_attn.qkv_proj.scales": "model.safetensors",
278
+ "model.layers.25.self_attn.qkv_proj.weight": "model.safetensors",
279
+ "model.layers.26.input_layernorm.weight": "model.safetensors",
280
+ "model.layers.26.mlp.down_proj.biases": "model.safetensors",
281
+ "model.layers.26.mlp.down_proj.scales": "model.safetensors",
282
+ "model.layers.26.mlp.down_proj.weight": "model.safetensors",
283
+ "model.layers.26.mlp.gate_up_proj.biases": "model.safetensors",
284
+ "model.layers.26.mlp.gate_up_proj.scales": "model.safetensors",
285
+ "model.layers.26.mlp.gate_up_proj.weight": "model.safetensors",
286
+ "model.layers.26.post_attention_layernorm.weight": "model.safetensors",
287
+ "model.layers.26.self_attn.o_proj.biases": "model.safetensors",
288
+ "model.layers.26.self_attn.o_proj.scales": "model.safetensors",
289
+ "model.layers.26.self_attn.o_proj.weight": "model.safetensors",
290
+ "model.layers.26.self_attn.qkv_proj.biases": "model.safetensors",
291
+ "model.layers.26.self_attn.qkv_proj.scales": "model.safetensors",
292
+ "model.layers.26.self_attn.qkv_proj.weight": "model.safetensors",
293
+ "model.layers.27.input_layernorm.weight": "model.safetensors",
294
+ "model.layers.27.mlp.down_proj.biases": "model.safetensors",
295
+ "model.layers.27.mlp.down_proj.scales": "model.safetensors",
296
+ "model.layers.27.mlp.down_proj.weight": "model.safetensors",
297
+ "model.layers.27.mlp.gate_up_proj.biases": "model.safetensors",
298
+ "model.layers.27.mlp.gate_up_proj.scales": "model.safetensors",
299
+ "model.layers.27.mlp.gate_up_proj.weight": "model.safetensors",
300
+ "model.layers.27.post_attention_layernorm.weight": "model.safetensors",
301
+ "model.layers.27.self_attn.o_proj.biases": "model.safetensors",
302
+ "model.layers.27.self_attn.o_proj.scales": "model.safetensors",
303
+ "model.layers.27.self_attn.o_proj.weight": "model.safetensors",
304
+ "model.layers.27.self_attn.qkv_proj.biases": "model.safetensors",
305
+ "model.layers.27.self_attn.qkv_proj.scales": "model.safetensors",
306
+ "model.layers.27.self_attn.qkv_proj.weight": "model.safetensors",
307
+ "model.layers.28.input_layernorm.weight": "model.safetensors",
308
+ "model.layers.28.mlp.down_proj.biases": "model.safetensors",
309
+ "model.layers.28.mlp.down_proj.scales": "model.safetensors",
310
+ "model.layers.28.mlp.down_proj.weight": "model.safetensors",
311
+ "model.layers.28.mlp.gate_up_proj.biases": "model.safetensors",
312
+ "model.layers.28.mlp.gate_up_proj.scales": "model.safetensors",
313
+ "model.layers.28.mlp.gate_up_proj.weight": "model.safetensors",
314
+ "model.layers.28.post_attention_layernorm.weight": "model.safetensors",
315
+ "model.layers.28.self_attn.o_proj.biases": "model.safetensors",
316
+ "model.layers.28.self_attn.o_proj.scales": "model.safetensors",
317
+ "model.layers.28.self_attn.o_proj.weight": "model.safetensors",
318
+ "model.layers.28.self_attn.qkv_proj.biases": "model.safetensors",
319
+ "model.layers.28.self_attn.qkv_proj.scales": "model.safetensors",
320
+ "model.layers.28.self_attn.qkv_proj.weight": "model.safetensors",
321
+ "model.layers.29.input_layernorm.weight": "model.safetensors",
322
+ "model.layers.29.mlp.down_proj.biases": "model.safetensors",
323
+ "model.layers.29.mlp.down_proj.scales": "model.safetensors",
324
+ "model.layers.29.mlp.down_proj.weight": "model.safetensors",
325
+ "model.layers.29.mlp.gate_up_proj.biases": "model.safetensors",
326
+ "model.layers.29.mlp.gate_up_proj.scales": "model.safetensors",
327
+ "model.layers.29.mlp.gate_up_proj.weight": "model.safetensors",
328
+ "model.layers.29.post_attention_layernorm.weight": "model.safetensors",
329
+ "model.layers.29.self_attn.o_proj.biases": "model.safetensors",
330
+ "model.layers.29.self_attn.o_proj.scales": "model.safetensors",
331
+ "model.layers.29.self_attn.o_proj.weight": "model.safetensors",
332
+ "model.layers.29.self_attn.qkv_proj.biases": "model.safetensors",
333
+ "model.layers.29.self_attn.qkv_proj.scales": "model.safetensors",
334
+ "model.layers.29.self_attn.qkv_proj.weight": "model.safetensors",
335
+ "model.layers.3.input_layernorm.weight": "model.safetensors",
336
+ "model.layers.3.mlp.down_proj.biases": "model.safetensors",
337
+ "model.layers.3.mlp.down_proj.scales": "model.safetensors",
338
+ "model.layers.3.mlp.down_proj.weight": "model.safetensors",
339
+ "model.layers.3.mlp.gate_up_proj.biases": "model.safetensors",
340
+ "model.layers.3.mlp.gate_up_proj.scales": "model.safetensors",
341
+ "model.layers.3.mlp.gate_up_proj.weight": "model.safetensors",
342
+ "model.layers.3.post_attention_layernorm.weight": "model.safetensors",
343
+ "model.layers.3.self_attn.o_proj.biases": "model.safetensors",
344
+ "model.layers.3.self_attn.o_proj.scales": "model.safetensors",
345
+ "model.layers.3.self_attn.o_proj.weight": "model.safetensors",
346
+ "model.layers.3.self_attn.qkv_proj.biases": "model.safetensors",
347
+ "model.layers.3.self_attn.qkv_proj.scales": "model.safetensors",
348
+ "model.layers.3.self_attn.qkv_proj.weight": "model.safetensors",
349
+ "model.layers.30.input_layernorm.weight": "model.safetensors",
350
+ "model.layers.30.mlp.down_proj.biases": "model.safetensors",
351
+ "model.layers.30.mlp.down_proj.scales": "model.safetensors",
352
+ "model.layers.30.mlp.down_proj.weight": "model.safetensors",
353
+ "model.layers.30.mlp.gate_up_proj.biases": "model.safetensors",
354
+ "model.layers.30.mlp.gate_up_proj.scales": "model.safetensors",
355
+ "model.layers.30.mlp.gate_up_proj.weight": "model.safetensors",
356
+ "model.layers.30.post_attention_layernorm.weight": "model.safetensors",
357
+ "model.layers.30.self_attn.o_proj.biases": "model.safetensors",
358
+ "model.layers.30.self_attn.o_proj.scales": "model.safetensors",
359
+ "model.layers.30.self_attn.o_proj.weight": "model.safetensors",
360
+ "model.layers.30.self_attn.qkv_proj.biases": "model.safetensors",
361
+ "model.layers.30.self_attn.qkv_proj.scales": "model.safetensors",
362
+ "model.layers.30.self_attn.qkv_proj.weight": "model.safetensors",
363
+ "model.layers.31.input_layernorm.weight": "model.safetensors",
364
+ "model.layers.31.mlp.down_proj.biases": "model.safetensors",
365
+ "model.layers.31.mlp.down_proj.scales": "model.safetensors",
366
+ "model.layers.31.mlp.down_proj.weight": "model.safetensors",
367
+ "model.layers.31.mlp.gate_up_proj.biases": "model.safetensors",
368
+ "model.layers.31.mlp.gate_up_proj.scales": "model.safetensors",
369
+ "model.layers.31.mlp.gate_up_proj.weight": "model.safetensors",
370
+ "model.layers.31.post_attention_layernorm.weight": "model.safetensors",
371
+ "model.layers.31.self_attn.o_proj.biases": "model.safetensors",
372
+ "model.layers.31.self_attn.o_proj.scales": "model.safetensors",
373
+ "model.layers.31.self_attn.o_proj.weight": "model.safetensors",
374
+ "model.layers.31.self_attn.qkv_proj.biases": "model.safetensors",
375
+ "model.layers.31.self_attn.qkv_proj.scales": "model.safetensors",
376
+ "model.layers.31.self_attn.qkv_proj.weight": "model.safetensors",
377
+ "model.layers.4.input_layernorm.weight": "model.safetensors",
378
+ "model.layers.4.mlp.down_proj.biases": "model.safetensors",
379
+ "model.layers.4.mlp.down_proj.scales": "model.safetensors",
380
+ "model.layers.4.mlp.down_proj.weight": "model.safetensors",
381
+ "model.layers.4.mlp.gate_up_proj.biases": "model.safetensors",
382
+ "model.layers.4.mlp.gate_up_proj.scales": "model.safetensors",
383
+ "model.layers.4.mlp.gate_up_proj.weight": "model.safetensors",
384
+ "model.layers.4.post_attention_layernorm.weight": "model.safetensors",
385
+ "model.layers.4.self_attn.o_proj.biases": "model.safetensors",
386
+ "model.layers.4.self_attn.o_proj.scales": "model.safetensors",
387
+ "model.layers.4.self_attn.o_proj.weight": "model.safetensors",
388
+ "model.layers.4.self_attn.qkv_proj.biases": "model.safetensors",
389
+ "model.layers.4.self_attn.qkv_proj.scales": "model.safetensors",
390
+ "model.layers.4.self_attn.qkv_proj.weight": "model.safetensors",
391
+ "model.layers.5.input_layernorm.weight": "model.safetensors",
392
+ "model.layers.5.mlp.down_proj.biases": "model.safetensors",
393
+ "model.layers.5.mlp.down_proj.scales": "model.safetensors",
394
+ "model.layers.5.mlp.down_proj.weight": "model.safetensors",
395
+ "model.layers.5.mlp.gate_up_proj.biases": "model.safetensors",
396
+ "model.layers.5.mlp.gate_up_proj.scales": "model.safetensors",
397
+ "model.layers.5.mlp.gate_up_proj.weight": "model.safetensors",
398
+ "model.layers.5.post_attention_layernorm.weight": "model.safetensors",
399
+ "model.layers.5.self_attn.o_proj.biases": "model.safetensors",
400
+ "model.layers.5.self_attn.o_proj.scales": "model.safetensors",
401
+ "model.layers.5.self_attn.o_proj.weight": "model.safetensors",
402
+ "model.layers.5.self_attn.qkv_proj.biases": "model.safetensors",
403
+ "model.layers.5.self_attn.qkv_proj.scales": "model.safetensors",
404
+ "model.layers.5.self_attn.qkv_proj.weight": "model.safetensors",
405
+ "model.layers.6.input_layernorm.weight": "model.safetensors",
406
+ "model.layers.6.mlp.down_proj.biases": "model.safetensors",
407
+ "model.layers.6.mlp.down_proj.scales": "model.safetensors",
408
+ "model.layers.6.mlp.down_proj.weight": "model.safetensors",
409
+ "model.layers.6.mlp.gate_up_proj.biases": "model.safetensors",
410
+ "model.layers.6.mlp.gate_up_proj.scales": "model.safetensors",
411
+ "model.layers.6.mlp.gate_up_proj.weight": "model.safetensors",
412
+ "model.layers.6.post_attention_layernorm.weight": "model.safetensors",
413
+ "model.layers.6.self_attn.o_proj.biases": "model.safetensors",
414
+ "model.layers.6.self_attn.o_proj.scales": "model.safetensors",
415
+ "model.layers.6.self_attn.o_proj.weight": "model.safetensors",
416
+ "model.layers.6.self_attn.qkv_proj.biases": "model.safetensors",
417
+ "model.layers.6.self_attn.qkv_proj.scales": "model.safetensors",
418
+ "model.layers.6.self_attn.qkv_proj.weight": "model.safetensors",
419
+ "model.layers.7.input_layernorm.weight": "model.safetensors",
420
+ "model.layers.7.mlp.down_proj.biases": "model.safetensors",
421
+ "model.layers.7.mlp.down_proj.scales": "model.safetensors",
422
+ "model.layers.7.mlp.down_proj.weight": "model.safetensors",
423
+ "model.layers.7.mlp.gate_up_proj.biases": "model.safetensors",
424
+ "model.layers.7.mlp.gate_up_proj.scales": "model.safetensors",
425
+ "model.layers.7.mlp.gate_up_proj.weight": "model.safetensors",
426
+ "model.layers.7.post_attention_layernorm.weight": "model.safetensors",
427
+ "model.layers.7.self_attn.o_proj.biases": "model.safetensors",
428
+ "model.layers.7.self_attn.o_proj.scales": "model.safetensors",
429
+ "model.layers.7.self_attn.o_proj.weight": "model.safetensors",
430
+ "model.layers.7.self_attn.qkv_proj.biases": "model.safetensors",
431
+ "model.layers.7.self_attn.qkv_proj.scales": "model.safetensors",
432
+ "model.layers.7.self_attn.qkv_proj.weight": "model.safetensors",
433
+ "model.layers.8.input_layernorm.weight": "model.safetensors",
434
+ "model.layers.8.mlp.down_proj.biases": "model.safetensors",
435
+ "model.layers.8.mlp.down_proj.scales": "model.safetensors",
436
+ "model.layers.8.mlp.down_proj.weight": "model.safetensors",
437
+ "model.layers.8.mlp.gate_up_proj.biases": "model.safetensors",
438
+ "model.layers.8.mlp.gate_up_proj.scales": "model.safetensors",
439
+ "model.layers.8.mlp.gate_up_proj.weight": "model.safetensors",
440
+ "model.layers.8.post_attention_layernorm.weight": "model.safetensors",
441
+ "model.layers.8.self_attn.o_proj.biases": "model.safetensors",
442
+ "model.layers.8.self_attn.o_proj.scales": "model.safetensors",
443
+ "model.layers.8.self_attn.o_proj.weight": "model.safetensors",
444
+ "model.layers.8.self_attn.qkv_proj.biases": "model.safetensors",
445
+ "model.layers.8.self_attn.qkv_proj.scales": "model.safetensors",
446
+ "model.layers.8.self_attn.qkv_proj.weight": "model.safetensors",
447
+ "model.layers.9.input_layernorm.weight": "model.safetensors",
448
+ "model.layers.9.mlp.down_proj.biases": "model.safetensors",
449
+ "model.layers.9.mlp.down_proj.scales": "model.safetensors",
450
+ "model.layers.9.mlp.down_proj.weight": "model.safetensors",
451
+ "model.layers.9.mlp.gate_up_proj.biases": "model.safetensors",
452
+ "model.layers.9.mlp.gate_up_proj.scales": "model.safetensors",
453
+ "model.layers.9.mlp.gate_up_proj.weight": "model.safetensors",
454
+ "model.layers.9.post_attention_layernorm.weight": "model.safetensors",
455
+ "model.layers.9.self_attn.o_proj.biases": "model.safetensors",
456
+ "model.layers.9.self_attn.o_proj.scales": "model.safetensors",
457
+ "model.layers.9.self_attn.o_proj.weight": "model.safetensors",
458
+ "model.layers.9.self_attn.qkv_proj.biases": "model.safetensors",
459
+ "model.layers.9.self_attn.qkv_proj.scales": "model.safetensors",
460
+ "model.layers.9.self_attn.qkv_proj.weight": "model.safetensors",
461
+ "model.norm.weight": "model.safetensors"
462
+ }
463
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": null,
3
+ "additional_special_tokens": null,
4
+ "backend": "tokenizers",
5
+ "bos_token": "<s>",
6
+ "clean_up_tokenization_spaces": false,
7
+ "eos_token": "<|endoftext|>",
8
+ "is_local": true,
9
+ "model_max_length": 131072,
10
+ "pad_token": "<|endoftext|>",
11
+ "padding_side": "left",
12
+ "sp_model_kwargs": {},
13
+ "tokenizer_class": "LlamaTokenizer",
14
+ "unk_token": "<unk>",
15
+ "use_default_system_prompt": false
16
+ }