Hugging Face's logo

Netsnake
/
Refact-1_6B-fim-Q5_K_M-GGUF

Text Generation
Transformers
GGUF
English
code
llama-cpp
gguf-my-repo
Eval Results (legacy)
Model card Files
xet
 Community
Netsnake commited on
Commit
2ecc144
·
verified ·
1 Parent(s): 6596b59

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +321 -0
README.md ADDED
@@ -0,0 +1,321 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ pipeline_tag: text-generation
3
+ inference: true
4
+ widget:
5
+ - text: 'def print_hello_world():'
6
+ example_title: Hello world
7
+ group: Python
8
+ license: bigscience-openrail-m
9
+ pretrain-datasets:
10
+ - books
11
+ - arxiv
12
+ - c4
13
+ - falcon-refinedweb
14
+ - wiki
15
+ - github-issues
16
+ - stack_markdown
17
+ - self-made dataset of permissive github code
18
+ datasets:
19
+ - bigcode/the-stack-dedup
20
+ - rombodawg/2XUNCENSORED_MegaCodeTraining188k
21
+ - bigcode/commitpackft
22
+ metrics:
23
+ - code_eval
24
+ library_name: transformers
25
+ tags:
26
+ - code
27
+ - llama-cpp
28
+ - gguf-my-repo
29
+ language:
30
+ - en
31
+ base_model: refactai/Refact-1_6B-fim
32
+ model-index:
33
+ - name: Refact-1.6B
34
+ results:
35
+ - task:
36
+ type: text-generation
37
+ dataset:
38
+ name: HumanEval
39
+ type: openai_humaneval
40
+ metrics:
41
+ - type: pass@1
42
+ value: 32.0
43
+ name: pass@1 (T=0.01)
44
+ verified: false
45
+ - type: pass@1
46
+ value: 31.5
47
+ name: pass@1 (T=0.2)
48
+ verified: false
49
+ - type: pass@10
50
+ value: 53.0
51
+ name: pass@10 (T=0.8)
52
+ verified: false
53
+ - type: pass@100
54
+ value: 76.9
55
+ name: pass@100 (T=0.8)
56
+ verified: false
57
+ - task:
58
+ type: text-generation
59
+ dataset:
60
+ name: HumanEvalSynthesize Python
61
+ type: bigcode/humanevalpack
62
+ metrics:
63
+ - type: pass@1
64
+ value: 35.8
65
+ name: pass@1 (T=0.2)
66
+ verified: false
67
+ - type: pass@1
68
+ value: 31.6
69
+ name: pass@1 (T=0.2)
70
+ verified: false
71
+ - type: pass@1
72
+ value: 29.1
73
+ name: pass@1 (T=0.2)
74
+ verified: false
75
+ - type: pass@1
76
+ value: -1
77
+ name: pass@1 (T=0.2)
78
+ verified: false
79
+ - type: pass@1
80
+ value: 26.3
81
+ name: pass@1 (T=0.2)
82
+ verified: false
83
+ - type: pass@1
84
+ value: -1
85
+ name: pass@1 (T=0.2)
86
+ verified: false
87
+ - type: pass@1
88
+ value: -1
89
+ name: pass@1 (T=0.2)
90
+ verified: false
91
+ - type: pass@1
92
+ value: 18.38
93
+ name: pass@1 (T=0.2)
94
+ verified: false
95
+ - type: pass@1
96
+ value: 12.28
97
+ name: pass@1 (T=0.2)
98
+ verified: false
99
+ - type: pass@1
100
+ value: 15.12
101
+ name: pass@1 (T=0.2)
102
+ verified: false
103
+ - type: pass@1
104
+ value: -1
105
+ name: pass@1 (T=0.2)
106
+ verified: false
107
+ - type: pass@1
108
+ value: 13.17
109
+ name: pass@1 (T=0.2)
110
+ verified: false
111
+ - type: pass@1
112
+ value: 2.8
113
+ name: pass@1 (T=0.2)
114
+ verified: false
115
+ - type: pass@1
116
+ value: -1
117
+ name: pass@1 (T=0.2)
118
+ verified: false
119
+ - type: pass@1
120
+ value: 26.92
121
+ name: pass@1 (T=0.2)
122
+ verified: false
123
+ - type: pass@1
124
+ value: 26.85
125
+ name: pass@1 (T=0.2)
126
+ verified: false
127
+ - type: pass@1
128
+ value: 30.76
129
+ name: pass@1 (T=0.2)
130
+ verified: false
131
+ - type: pass@1
132
+ value: -1
133
+ name: pass@1 (T=0.2)
134
+ verified: false
135
+ - type: pass@1
136
+ value: 25.94
137
+ name: pass@1 (T=0.2)
138
+ verified: false
139
+ - type: pass@1
140
+ value: 8.44
141
+ name: pass@1 (T=0.2)
142
+ verified: false
143
+ - type: pass@1
144
+ value: -1
145
+ name: pass@1 (T=0.2)
146
+ verified: false
147
+ - type: pass@1
148
+ value: 26.46
149
+ name: pass@1 (T=0.2)
150
+ verified: false
151
+ - type: pass@1
152
+ value: 17.86
153
+ name: pass@1 (T=0.2)
154
+ verified: false
155
+ - type: pass@1
156
+ value: 20.94
157
+ name: pass@1 (T=0.2)
158
+ verified: false
159
+ - type: pass@1
160
+ value: -1
161
+ name: pass@1 (T=0.2)
162
+ verified: false
163
+ - type: pass@1
164
+ value: 18.78
165
+ name: pass@1 (T=0.2)
166
+ verified: false
167
+ - type: pass@1
168
+ value: -1
169
+ name: pass@1 (T=0.2)
170
+ verified: false
171
+ - type: pass@1
172
+ value: -1
173
+ name: pass@1 (T=0.2)
174
+ verified: false
175
+ - task:
176
+ type: text-generation
177
+ dataset:
178
+ name: MBPP
179
+ type: mbpp
180
+ metrics:
181
+ - type: pass@1
182
+ value: 31.15
183
+ name: pass@1 (T=0.01)
184
+ verified: false
185
+ - task:
186
+ type: text-generation
187
+ dataset:
188
+ name: DS-1000 (Overall Completion)
189
+ type: ds1000
190
+ metrics:
191
+ - type: pass@1
192
+ value: 10.1
193
+ name: pass@1 (T=0.2)
194
+ verified: false
195
+ - task:
196
+ type: text-generation
197
+ dataset:
198
+ name: MultiPL-HumanEval (C++)
199
+ type: nuprl/MultiPL-E
200
+ metrics:
201
+ - type: pass@1
202
+ value: 21.61
203
+ name: pass@1 (T=0.2)
204
+ verified: false
205
+ - type: pass@1
206
+ value: 13.91
207
+ name: pass@1 (T=0.2)
208
+ verified: false
209
+ - type: pass@1
210
+ value: 9.5
211
+ name: pass@1 (T=0.2)
212
+ verified: false
213
+ - type: pass@1
214
+ value: 53.57
215
+ name: pass@1 (T=0.2)
216
+ verified: false
217
+ - type: pass@1
218
+ value: 21.58
219
+ name: pass@1 (T=0.2)
220
+ verified: false
221
+ - type: pass@1
222
+ value: 13.75
223
+ name: pass@1 (T=0.2)
224
+ verified: false
225
+ - type: pass@1
226
+ value: 26.88
227
+ name: pass@1 (T=0.2)
228
+ verified: false
229
+ - type: pass@1
230
+ value: 15.26
231
+ name: pass@1 (T=0.2)
232
+ verified: false
233
+ - type: pass@1
234
+ value: 23.04
235
+ name: pass@1 (T=0.2)
236
+ verified: false
237
+ - type: pass@1
238
+ value: 12.1
239
+ name: pass@1 (T=0.2)
240
+ verified: false
241
+ - type: pass@1
242
+ value: 29.6
243
+ name: pass@1 (T=0.2)
244
+ verified: false
245
+ - type: pass@1
246
+ value: 13.77
247
+ name: pass@1 (T=0.2)
248
+ verified: false
249
+ - type: pass@1
250
+ value: 12.68
251
+ name: pass@1 (T=0.2)
252
+ verified: false
253
+ - type: pass@1
254
+ value: 4.29
255
+ name: pass@1 (T=0.2)
256
+ verified: false
257
+ - type: pass@1
258
+ value: 19.54
259
+ name: pass@1 (T=0.2)
260
+ verified: false
261
+ - type: pass@1
262
+ value: 18.33
263
+ name: pass@1 (T=0.2)
264
+ verified: false
265
+ - type: pass@1
266
+ value: 5.7
267
+ name: pass@1 (T=0.2)
268
+ verified: false
269
+ - type: pass@1
270
+ value: 17.68
271
+ name: pass@1 (T=0.2)
272
+ verified: false
273
+ - type: pass@1
274
+ value: 25
275
+ name: pass@1 (T=0.2)
276
+ verified: false
277
+ ---
278
+
279
+ # Netsnake/Refact-1_6B-fim-Q5_K_M-GGUF
280
+ This model was converted to GGUF format from [`refactai/Refact-1_6B-fim`](https://huggingface.co/refactai/Refact-1_6B-fim) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
281
+ Refer to the [original model card](https://huggingface.co/refactai/Refact-1_6B-fim) for more details on the model.
282
+
283
+ ## Use with llama.cpp
284
+ Install llama.cpp through brew (works on Mac and Linux)
285
+
286
+ ```bash
287
+ brew install llama.cpp
288
+
289
+ ```
290
+ Invoke the llama.cpp server or the CLI.
291
+
292
+ ### CLI:
293
+ ```bash
294
+ llama-cli --hf-repo Netsnake/Refact-1_6B-fim-Q5_K_M-GGUF --hf-file refact-1_6b-fim-q5_k_m.gguf -p "The meaning to life and the universe is"
295
+ ```
296
+
297
+ ### Server:
298
+ ```bash
299
+ llama-server --hf-repo Netsnake/Refact-1_6B-fim-Q5_K_M-GGUF --hf-file refact-1_6b-fim-q5_k_m.gguf -c 2048
300
+ ```
301
+
302
+ Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well.
303
+
304
+ Step 1: Clone llama.cpp from GitHub.
305
+ ```
306
+ git clone https://github.com/ggerganov/llama.cpp
307
+ ```
308
+
309
+ Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).
310
+ ```
311
+ cd llama.cpp && LLAMA_CURL=1 make
312
+ ```
313
+
314
+ Step 3: Run inference through the main binary.
315
+ ```
316
+ ./llama-cli --hf-repo Netsnake/Refact-1_6B-fim-Q5_K_M-GGUF --hf-file refact-1_6b-fim-q5_k_m.gguf -p "The meaning to life and the universe is"
317
+ ```
318
+ or
319
+ ```
320
+ ./llama-server --hf-repo Netsnake/Refact-1_6B-fim-Q5_K_M-GGUF --hf-file refact-1_6b-fim-q5_k_m.gguf -c 2048
321
+ ```