Hugging Face's logo

RichardErkhov
/
HuggingFaceH4_-_zephyr-7b-beta-gguf

GGUF
Inference Endpoints
conversational
Model card Files Community
RichardErkhov commited on
Commit
0c119ad
•
1 Parent(s): fe48f84

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +464 -0
README.md ADDED
@@ -0,0 +1,464 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Quantization made by Richard Erkhov.
2
+
3
+ [Github](https://github.com/RichardErkhov)
4
+
5
+ [Discord](https://discord.gg/pvy7H8DZMG)
6
+
7
+ [Request more models](https://github.com/RichardErkhov/quant_request)
8
+
9
+
10
+ zephyr-7b-beta - GGUF
11
+ - Model creator: https://huggingface.co/HuggingFaceH4/
12
+ - Original model: https://huggingface.co/HuggingFaceH4/zephyr-7b-beta/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [zephyr-7b-beta.Q2_K.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q2_K.gguf) | Q2_K | 2.53GB |
18
+ | [zephyr-7b-beta.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.IQ3_XS.gguf) | IQ3_XS | 2.81GB |
19
+ | [zephyr-7b-beta.IQ3_S.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.IQ3_S.gguf) | IQ3_S | 2.96GB |
20
+ | [zephyr-7b-beta.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q3_K_S.gguf) | Q3_K_S | 2.95GB |
21
+ | [zephyr-7b-beta.IQ3_M.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.IQ3_M.gguf) | IQ3_M | 3.06GB |
22
+ | [zephyr-7b-beta.Q3_K.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q3_K.gguf) | Q3_K | 3.28GB |
23
+ | [zephyr-7b-beta.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q3_K_M.gguf) | Q3_K_M | 3.28GB |
24
+ | [zephyr-7b-beta.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q3_K_L.gguf) | Q3_K_L | 3.56GB |
25
+ | [zephyr-7b-beta.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.IQ4_XS.gguf) | IQ4_XS | 3.67GB |
26
+ | [zephyr-7b-beta.Q4_0.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q4_0.gguf) | Q4_0 | 3.83GB |
27
+ | [zephyr-7b-beta.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.IQ4_NL.gguf) | IQ4_NL | 3.87GB |
28
+ | [zephyr-7b-beta.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q4_K_S.gguf) | Q4_K_S | 3.86GB |
29
+ | [zephyr-7b-beta.Q4_K.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q4_K.gguf) | Q4_K | 4.07GB |
30
+ | [zephyr-7b-beta.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q4_K_M.gguf) | Q4_K_M | 4.07GB |
31
+ | [zephyr-7b-beta.Q4_1.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q4_1.gguf) | Q4_1 | 4.24GB |
32
+ | [zephyr-7b-beta.Q5_0.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q5_0.gguf) | Q5_0 | 4.65GB |
33
+ | [zephyr-7b-beta.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q5_K_S.gguf) | Q5_K_S | 4.65GB |
34
+ | [zephyr-7b-beta.Q5_K.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q5_K.gguf) | Q5_K | 4.78GB |
35
+ | [zephyr-7b-beta.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q5_K_M.gguf) | Q5_K_M | 4.78GB |
36
+ | [zephyr-7b-beta.Q5_1.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q5_1.gguf) | Q5_1 | 5.07GB |
37
+ | [zephyr-7b-beta.Q6_K.gguf](https://huggingface.co/RichardErkhov/HuggingFaceH4_-_zephyr-7b-beta-gguf/blob/main/zephyr-7b-beta.Q6_K.gguf) | Q6_K | 5.53GB |
38
+
39
+
40
+
41
+
42
+ Original model description:
43
+ ---
44
+ tags:
45
+ - generated_from_trainer
46
+ license: mit
47
+ datasets:
48
+ - HuggingFaceH4/ultrachat_200k
49
+ - HuggingFaceH4/ultrafeedback_binarized
50
+ language:
51
+ - en
52
+ base_model: mistralai/Mistral-7B-v0.1
53
+ widget:
54
+ - example_title: Pirate!
55
+ messages:
56
+ - role: system
57
+ content: You are a pirate chatbot who always responds with Arr!
58
+ - role: user
59
+ content: "There's a llama on my lawn, how can I get rid of him?"
60
+ output:
61
+ text: >-
62
+ Arr! 'Tis a puzzlin' matter, me hearty! A llama on yer lawn be a rare
63
+ sight, but I've got a plan that might help ye get rid of 'im. Ye'll need
64
+ to gather some carrots and hay, and then lure the llama away with the
65
+ promise of a tasty treat. Once he's gone, ye can clean up yer lawn and
66
+ enjoy the peace and quiet once again. But beware, me hearty, for there
67
+ may be more llamas where that one came from! Arr!
68
+ pipeline_tag: text-generation
69
+ model-index:
70
+ - name: zephyr-7b-beta
71
+ results:
72
+ # AI2 Reasoning Challenge (25-Shot)
73
+ - task:
74
+ type: text-generation
75
+ name: Text Generation
76
+ dataset:
77
+ name: AI2 Reasoning Challenge (25-Shot)
78
+ type: ai2_arc
79
+ config: ARC-Challenge
80
+ split: test
81
+ args:
82
+ num_few_shot: 25
83
+ metrics:
84
+ - type: acc_norm
85
+ name: normalized accuracy
86
+ value: 62.03071672354948
87
+ source:
88
+ name: Open LLM Leaderboard
89
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
90
+
91
+ # HellaSwag (10-shot)
92
+ - task:
93
+ type: text-generation
94
+ name: Text Generation
95
+ dataset:
96
+ name: HellaSwag (10-Shot)
97
+ type: hellaswag
98
+ split: validation
99
+ args:
100
+ num_few_shot: 10
101
+ metrics:
102
+ - type: acc_norm
103
+ name: normalized accuracy
104
+ value: 84.35570603465445
105
+ source:
106
+ name: Open LLM Leaderboard
107
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
108
+
109
+ # DROP (3-shot)
110
+ - task:
111
+ type: text-generation
112
+ name: Text Generation
113
+ dataset:
114
+ name: Drop (3-Shot)
115
+ type: drop
116
+ split: validation
117
+ args:
118
+ num_few_shot: 3
119
+ metrics:
120
+ - type: f1
121
+ name: f1 score
122
+ value: 9.662437080536909
123
+ source:
124
+ name: Open LLM Leaderboard
125
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
126
+
127
+ # TruthfulQA (0-shot)
128
+ - task:
129
+ type: text-generation
130
+ name: Text Generation
131
+ dataset:
132
+ name: TruthfulQA (0-shot)
133
+ type: truthful_qa
134
+ config: multiple_choice
135
+ split: validation
136
+ args:
137
+ num_few_shot: 0
138
+ metrics:
139
+ - type: mc2
140
+ value: 57.44916942762855
141
+ source:
142
+ name: Open LLM Leaderboard
143
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
144
+
145
+ # GSM8k (5-shot)
146
+ - task:
147
+ type: text-generation
148
+ name: Text Generation
149
+ dataset:
150
+ name: GSM8k (5-shot)
151
+ type: gsm8k
152
+ config: main
153
+ split: test
154
+ args:
155
+ num_few_shot: 5
156
+ metrics:
157
+ - type: acc
158
+ name: accuracy
159
+ value: 12.736921910538287
160
+ source:
161
+ name: Open LLM Leaderboard
162
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
163
+
164
+ # MMLU (5-Shot)
165
+ - task:
166
+ type: text-generation
167
+ name: Text Generation
168
+ dataset:
169
+ name: MMLU (5-Shot)
170
+ type: cais/mmlu
171
+ config: all
172
+ split: test
173
+ args:
174
+ num_few_shot: 5
175
+ metrics:
176
+ - type: acc
177
+ name: accuracy
178
+ value: 61.07
179
+ source:
180
+ name: Open LLM Leaderboard
181
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
182
+
183
+ # Winogrande (5-shot)
184
+ - task:
185
+ type: text-generation
186
+ name: Text Generation
187
+ dataset:
188
+ name: Winogrande (5-shot)
189
+ type: winogrande
190
+ config: winogrande_xl
191
+ split: validation
192
+ args:
193
+ num_few_shot: 5
194
+ metrics:
195
+ - type: acc
196
+ name: accuracy
197
+ value: 77.74269928966061
198
+ source:
199
+ name: Open LLM Leaderboard
200
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=HuggingFaceH4/zephyr-7b-beta
201
+
202
+ # AlpacaEval (taken from model card)
203
+ - task:
204
+ type: text-generation
205
+ name: Text Generation
206
+ dataset:
207
+ name: AlpacaEval
208
+ type: tatsu-lab/alpaca_eval
209
+ metrics:
210
+ - type: unknown
211
+ name: win rate
212
+ value: 0.9060
213
+ source:
214
+ url: https://tatsu-lab.github.io/alpaca_eval/
215
+
216
+ # MT-Bench (taken from model card)
217
+ - task:
218
+ type: text-generation
219
+ name: Text Generation
220
+ dataset:
221
+ name: MT-Bench
222
+ type: unknown
223
+ metrics:
224
+ - type: unknown
225
+ name: score
226
+ value: 7.34
227
+ source:
228
+ url: https://huggingface.co/spaces/lmsys/mt-bench
229
+ ---
230
+
231
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
232
+ should probably proofread and complete it, then remove this comment. -->
233
+
234
+ <img src="https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha/resolve/main/thumbnail.png" alt="Zephyr Logo" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>
235
+
236
+
237
+ # Model Card for Zephyr 7B β
238
+
239
+ Zephyr is a series of language models that are trained to act as helpful assistants. Zephyr-7B-β is the second model in the series, and is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) that was trained on on a mix of publicly available, synthetic datasets using [Direct Preference Optimization (DPO)](https://arxiv.org/abs/2305.18290). We found that removing the in-built alignment of these datasets boosted performance on [MT Bench](https://huggingface.co/spaces/lmsys/mt-bench) and made the model more helpful. However, this means that model is likely to generate problematic text when prompted to do so. You can find more details in the [technical report](https://arxiv.org/abs/2310.16944).
240
+
241
+
242
+ ## Model description
243
+
244
+ - **Model type:** A 7B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets.
245
+ - **Language(s) (NLP):** Primarily English
246
+ - **License:** MIT
247
+ - **Finetuned from model:** [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
248
+
249
+ ### Model Sources
250
+
251
+ <!-- Provide the basic links for the model. -->
252
+
253
+ - **Repository:** https://github.com/huggingface/alignment-handbook
254
+ - **Demo:** https://huggingface.co/spaces/HuggingFaceH4/zephyr-chat
255
+ - **Chatbot Arena:** Evaluate Zephyr 7B against 10+ LLMs in the LMSYS arena: http://arena.lmsys.org
256
+
257
+ ## Performance
258
+
259
+ At the time of release, Zephyr-7B-β is the highest ranked 7B chat model on the [MT-Bench](https://huggingface.co/spaces/lmsys/mt-bench) and [AlpacaEval](https://tatsu-lab.github.io/alpaca_eval/) benchmarks:
260
+
261
+ | Model | Size | Alignment | MT-Bench (score) | AlpacaEval (win rate %) |
262
+ |-------------|-----|----|---------------|--------------|
263
+ | StableLM-Tuned-α | 7B| dSFT |2.75| -|
264
+ | MPT-Chat | 7B |dSFT |5.42| -|
265
+ | Xwin-LMv0.1 | 7B| dPPO| 6.19| 87.83|
266
+ | Mistral-Instructv0.1 | 7B| - | 6.84 |-|
267
+ | Zephyr-7b-α |7B| dDPO| 6.88| -|
268
+ | **Zephyr-7b-β** 🪁 | **7B** | **dDPO** | **7.34** | **90.60** |
269
+ | Falcon-Instruct | 40B |dSFT |5.17 |45.71|
270
+ | Guanaco | 65B | SFT |6.41| 71.80|
271
+ | Llama2-Chat | 70B |RLHF |6.86| 92.66|
272
+ | Vicuna v1.3 | 33B |dSFT |7.12 |88.99|
273
+ | WizardLM v1.0 | 70B |dSFT |7.71 |-|
274
+ | Xwin-LM v0.1 | 70B |dPPO |- |95.57|
275
+ | GPT-3.5-turbo | - |RLHF |7.94 |89.37|
276
+ | Claude 2 | - |RLHF |8.06| 91.36|
277
+ | GPT-4 | -| RLHF |8.99| 95.28|
278
+
279
+ In particular, on several categories of MT-Bench, Zephyr-7B-β has strong performance compared to larger open models like Llama2-Chat-70B:
280
+
281
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6200d0a443eb0913fa2df7cc/raxvt5ma16d7T23my34WC.png)
282
+
283
+ However, on more complex tasks like coding and mathematics, Zephyr-7B-β lags behind proprietary models and more research is needed to close the gap.
284
+
285
+
286
+ ## Intended uses & limitations
287
+
288
+ The model was initially fine-tuned on a filtered and preprocessed of the [`UltraChat`](https://huggingface.co/datasets/stingning/ultrachat) dataset, which contains a diverse range of synthetic dialogues generated by ChatGPT.
289
+ We then further aligned the model with [🤗 TRL's](https://github.com/huggingface/trl) `DPOTrainer` on the [openbmb/UltraFeedback](https://huggingface.co/datasets/openbmb/UltraFeedback) dataset, which contains 64k prompts and model completions that are ranked by GPT-4. As a result, the model can be used for chat and you can check out our [demo](https://huggingface.co/spaces/HuggingFaceH4/zephyr-chat) to test its capabilities.
290
+
291
+ You can find the datasets used for training Zephyr-7B-β [here](https://huggingface.co/collections/HuggingFaceH4/zephyr-7b-6538c6d6d5ddd1cbb1744a66)
292
+
293
+ Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
294
+
295
+ ```python
296
+ # Install transformers from source - only needed for versions <= v4.34
297
+ # pip install git+https://github.com/huggingface/transformers.git
298
+ # pip install accelerate
299
+
300
+ import torch
301
+ from transformers import pipeline
302
+
303
+ pipe = pipeline("text-generation", model="HuggingFaceH4/zephyr-7b-beta", torch_dtype=torch.bfloat16, device_map="auto")
304
+
305
+ # We use the tokenizer's chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
306
+ messages = [
307
+ {
308
+ "role": "system",
309
+ "content": "You are a friendly chatbot who always responds in the style of a pirate",
310
+ },
311
+ {"role": "user", "content": "How many helicopters can a human eat in one sitting?"},
312
+ ]
313
+ prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
314
+ outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
315
+ print(outputs[0]["generated_text"])
316
+ # <|system|>
317
+ # You are a friendly chatbot who always responds in the style of a pirate.</s>
318
+ # <|user|>
319
+ # How many helicopters can a human eat in one sitting?</s>
320
+ # <|assistant|>
321
+ # Ah, me hearty matey! But yer question be a puzzler! A human cannot eat a helicopter in one sitting, as helicopters are not edible. They be made of metal, plastic, and other materials, not food!
322
+ ```
323
+
324
+ ## Bias, Risks, and Limitations
325
+
326
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
327
+
328
+ Zephyr-7B-β has not been aligned to human preferences for safety within the RLHF phase or deployed with in-the-loop filtering of responses like ChatGPT, so the model can produce problematic outputs (especially when prompted to do so).
329
+ It is also unknown what the size and composition of the corpus was used to train the base model (`mistralai/Mistral-7B-v0.1`), however it is likely to have included a mix of Web data and technical sources like books and code. See the [Falcon 180B model card](https://huggingface.co/tiiuae/falcon-180B#training-data) for an example of this.
330
+
331
+
332
+ ## Training and evaluation data
333
+
334
+ During DPO training, this model achieves the following results on the evaluation set:
335
+
336
+ - Loss: 0.7496
337
+ - Rewards/chosen: -4.5221
338
+ - Rewards/rejected: -8.3184
339
+ - Rewards/accuracies: 0.7812
340
+ - Rewards/margins: 3.7963
341
+ - Logps/rejected: -340.1541
342
+ - Logps/chosen: -299.4561
343
+ - Logits/rejected: -2.3081
344
+ - Logits/chosen: -2.3531
345
+
346
+
347
+ ### Training hyperparameters
348
+
349
+ The following hyperparameters were used during training:
350
+ - learning_rate: 5e-07
351
+ - train_batch_size: 2
352
+ - eval_batch_size: 4
353
+ - seed: 42
354
+ - distributed_type: multi-GPU
355
+ - num_devices: 16
356
+ - total_train_batch_size: 32
357
+ - total_eval_batch_size: 64
358
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
359
+ - lr_scheduler_type: linear
360
+ - lr_scheduler_warmup_ratio: 0.1
361
+ - num_epochs: 3.0
362
+
363
+ ### Training results
364
+
365
+ The table below shows the full set of DPO training metrics:
366
+
367
+
368
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
369
+ |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
370
+ | 0.6284 | 0.05 | 100 | 0.6098 | 0.0425 | -0.1872 | 0.7344 | 0.2297 | -258.8416 | -253.8099 | -2.7976 | -2.8234 |
371
+ | 0.4908 | 0.1 | 200 | 0.5426 | -0.0279 | -0.6842 | 0.75 | 0.6563 | -263.8124 | -254.5145 | -2.7719 | -2.7960 |
372
+ | 0.5264 | 0.15 | 300 | 0.5324 | 0.0414 | -0.9793 | 0.7656 | 1.0207 | -266.7627 | -253.8209 | -2.7892 | -2.8122 |
373
+ | 0.5536 | 0.21 | 400 | 0.4957 | -0.0185 | -1.5276 | 0.7969 | 1.5091 | -272.2460 | -254.4203 | -2.8542 | -2.8764 |
374
+ | 0.5362 | 0.26 | 500 | 0.5031 | -0.2630 | -1.5917 | 0.7812 | 1.3287 | -272.8869 | -256.8653 | -2.8702 | -2.8958 |
375
+ | 0.5966 | 0.31 | 600 | 0.5963 | -0.2993 | -1.6491 | 0.7812 | 1.3499 | -273.4614 | -257.2279 | -2.8778 | -2.8986 |
376
+ | 0.5014 | 0.36 | 700 | 0.5382 | -0.2859 | -1.4750 | 0.75 | 1.1891 | -271.7204 | -257.0942 | -2.7659 | -2.7869 |
377
+ | 0.5334 | 0.41 | 800 | 0.5677 | -0.4289 | -1.8968 | 0.7969 | 1.4679 | -275.9378 | -258.5242 | -2.7053 | -2.7265 |
378
+ | 0.5251 | 0.46 | 900 | 0.5772 | -0.2116 | -1.3107 | 0.7344 | 1.0991 | -270.0768 | -256.3507 | -2.8463 | -2.8662 |
379
+ | 0.5205 | 0.52 | 1000 | 0.5262 | -0.3792 | -1.8585 | 0.7188 | 1.4793 | -275.5552 | -258.0276 | -2.7893 | -2.7979 |
380
+ | 0.5094 | 0.57 | 1100 | 0.5433 | -0.6279 | -1.9368 | 0.7969 | 1.3089 | -276.3377 | -260.5136 | -2.7453 | -2.7536 |
381
+ | 0.5837 | 0.62 | 1200 | 0.5349 | -0.3780 | -1.9584 | 0.7656 | 1.5804 | -276.5542 | -258.0154 | -2.7643 | -2.7756 |
382
+ | 0.5214 | 0.67 | 1300 | 0.5732 | -1.0055 | -2.2306 | 0.7656 | 1.2251 | -279.2761 | -264.2903 | -2.6986 | -2.7113 |
383
+ | 0.6914 | 0.72 | 1400 | 0.5137 | -0.6912 | -2.1775 | 0.7969 | 1.4863 | -278.7448 | -261.1467 | -2.7166 | -2.7275 |
384
+ | 0.4655 | 0.77 | 1500 | 0.5090 | -0.7987 | -2.2930 | 0.7031 | 1.4943 | -279.8999 | -262.2220 | -2.6651 | -2.6838 |
385
+ | 0.5731 | 0.83 | 1600 | 0.5312 | -0.8253 | -2.3520 | 0.7812 | 1.5268 | -280.4902 | -262.4876 | -2.6543 | -2.6728 |
386
+ | 0.5233 | 0.88 | 1700 | 0.5206 | -0.4573 | -2.0951 | 0.7812 | 1.6377 | -277.9205 | -258.8084 | -2.6870 | -2.7097 |
387
+ | 0.5593 | 0.93 | 1800 | 0.5231 | -0.5508 | -2.2000 | 0.7969 | 1.6492 | -278.9703 | -259.7433 | -2.6221 | -2.6519 |
388
+ | 0.4967 | 0.98 | 1900 | 0.5290 | -0.5340 | -1.9570 | 0.8281 | 1.4230 | -276.5395 | -259.5749 | -2.6564 | -2.6878 |
389
+ | 0.0921 | 1.03 | 2000 | 0.5368 | -1.1376 | -3.1615 | 0.7812 | 2.0239 | -288.5854 | -265.6111 | -2.6040 | -2.6345 |
390
+ | 0.0733 | 1.08 | 2100 | 0.5453 | -1.1045 | -3.4451 | 0.7656 | 2.3406 | -291.4208 | -265.2799 | -2.6289 | -2.6595 |
391
+ | 0.0972 | 1.14 | 2200 | 0.5571 | -1.6915 | -3.9823 | 0.8125 | 2.2908 | -296.7934 | -271.1505 | -2.6471 | -2.6709 |
392
+ | 0.1058 | 1.19 | 2300 | 0.5789 | -1.0621 | -3.8941 | 0.7969 | 2.8319 | -295.9106 | -264.8563 | -2.5527 | -2.5798 |
393
+ | 0.2423 | 1.24 | 2400 | 0.5455 | -1.1963 | -3.5590 | 0.7812 | 2.3627 | -292.5599 | -266.1981 | -2.5414 | -2.5784 |
394
+ | 0.1177 | 1.29 | 2500 | 0.5889 | -1.8141 | -4.3942 | 0.7969 | 2.5801 | -300.9120 | -272.3761 | -2.4802 | -2.5189 |
395
+ | 0.1213 | 1.34 | 2600 | 0.5683 | -1.4608 | -3.8420 | 0.8125 | 2.3812 | -295.3901 | -268.8436 | -2.4774 | -2.5207 |
396
+ | 0.0889 | 1.39 | 2700 | 0.5890 | -1.6007 | -3.7337 | 0.7812 | 2.1330 | -294.3068 | -270.2423 | -2.4123 | -2.4522 |
397
+ | 0.0995 | 1.45 | 2800 | 0.6073 | -1.5519 | -3.8362 | 0.8281 | 2.2843 | -295.3315 | -269.7538 | -2.4685 | -2.5050 |
398
+ | 0.1145 | 1.5 | 2900 | 0.5790 | -1.7939 | -4.2876 | 0.8438 | 2.4937 | -299.8461 | -272.1744 | -2.4272 | -2.4674 |
399
+ | 0.0644 | 1.55 | 3000 | 0.5735 | -1.7285 | -4.2051 | 0.8125 | 2.4766 | -299.0209 | -271.5201 | -2.4193 | -2.4574 |
400
+ | 0.0798 | 1.6 | 3100 | 0.5537 | -1.7226 | -4.2850 | 0.8438 | 2.5624 | -299.8200 | -271.4610 | -2.5367 | -2.5696 |
401
+ | 0.1013 | 1.65 | 3200 | 0.5575 | -1.5715 | -3.9813 | 0.875 | 2.4098 | -296.7825 | -269.9498 | -2.4926 | -2.5267 |
402
+ | 0.1254 | 1.7 | 3300 | 0.5905 | -1.6412 | -4.4703 | 0.8594 | 2.8291 | -301.6730 | -270.6473 | -2.5017 | -2.5340 |
403
+ | 0.085 | 1.76 | 3400 | 0.6133 | -1.9159 | -4.6760 | 0.8438 | 2.7601 | -303.7296 | -273.3941 | -2.4614 | -2.4960 |
404
+ | 0.065 | 1.81 | 3500 | 0.6074 | -1.8237 | -4.3525 | 0.8594 | 2.5288 | -300.4951 | -272.4724 | -2.4597 | -2.5004 |
405
+ | 0.0755 | 1.86 | 3600 | 0.5836 | -1.9252 | -4.4005 | 0.8125 | 2.4753 | -300.9748 | -273.4872 | -2.4327 | -2.4716 |
406
+ | 0.0746 | 1.91 | 3700 | 0.5789 | -1.9280 | -4.4906 | 0.8125 | 2.5626 | -301.8762 | -273.5149 | -2.4686 | -2.5115 |
407
+ | 0.1348 | 1.96 | 3800 | 0.6015 | -1.8658 | -4.2428 | 0.8281 | 2.3769 | -299.3976 | -272.8936 | -2.4943 | -2.5393 |
408
+ | 0.0217 | 2.01 | 3900 | 0.6122 | -2.3335 | -4.9229 | 0.8281 | 2.5894 | -306.1988 | -277.5699 | -2.4841 | -2.5272 |
409
+ | 0.0219 | 2.07 | 4000 | 0.6522 | -2.9890 | -6.0164 | 0.8281 | 3.0274 | -317.1334 | -284.1248 | -2.4105 | -2.4545 |
410
+ | 0.0119 | 2.12 | 4100 | 0.6922 | -3.4777 | -6.6749 | 0.7969 | 3.1972 | -323.7187 | -289.0121 | -2.4272 | -2.4699 |
411
+ | 0.0153 | 2.17 | 4200 | 0.6993 | -3.2406 | -6.6775 | 0.7969 | 3.4369 | -323.7453 | -286.6413 | -2.4047 | -2.4465 |
412
+ | 0.011 | 2.22 | 4300 | 0.7178 | -3.7991 | -7.4397 | 0.7656 | 3.6406 | -331.3667 | -292.2260 | -2.3843 | -2.4290 |
413
+ | 0.0072 | 2.27 | 4400 | 0.6840 | -3.3269 | -6.8021 | 0.8125 | 3.4752 | -324.9908 | -287.5042 | -2.4095 | -2.4536 |
414
+ | 0.0197 | 2.32 | 4500 | 0.7013 | -3.6890 | -7.3014 | 0.8125 | 3.6124 | -329.9841 | -291.1250 | -2.4118 | -2.4543 |
415
+ | 0.0182 | 2.37 | 4600 | 0.7476 | -3.8994 | -7.5366 | 0.8281 | 3.6372 | -332.3356 | -293.2291 | -2.4163 | -2.4565 |
416
+ | 0.0125 | 2.43 | 4700 | 0.7199 | -4.0560 | -7.5765 | 0.8438 | 3.5204 | -332.7345 | -294.7952 | -2.3699 | -2.4100 |
417
+ | 0.0082 | 2.48 | 4800 | 0.7048 | -3.6613 | -7.1356 | 0.875 | 3.4743 | -328.3255 | -290.8477 | -2.3925 | -2.4303 |
418
+ | 0.0118 | 2.53 | 4900 | 0.6976 | -3.7908 | -7.3152 | 0.8125 | 3.5244 | -330.1224 | -292.1431 | -2.3633 | -2.4047 |
419
+ | 0.0118 | 2.58 | 5000 | 0.7198 | -3.9049 | -7.5557 | 0.8281 | 3.6508 | -332.5271 | -293.2844 | -2.3764 | -2.4194 |
420
+ | 0.006 | 2.63 | 5100 | 0.7506 | -4.2118 | -7.9149 | 0.8125 | 3.7032 | -336.1194 | -296.3530 | -2.3407 | -2.3860 |
421
+ | 0.0143 | 2.68 | 5200 | 0.7408 | -4.2433 | -7.9802 | 0.8125 | 3.7369 | -336.7721 | -296.6682 | -2.3509 | -2.3946 |
422
+ | 0.0057 | 2.74 | 5300 | 0.7552 | -4.3392 | -8.0831 | 0.7969 | 3.7439 | -337.8013 | -297.6275 | -2.3388 | -2.3842 |
423
+ | 0.0138 | 2.79 | 5400 | 0.7404 | -4.2395 | -7.9762 | 0.8125 | 3.7367 | -336.7322 | -296.6304 | -2.3286 | -2.3737 |
424
+ | 0.0079 | 2.84 | 5500 | 0.7525 | -4.4466 | -8.2196 | 0.7812 | 3.7731 | -339.1662 | -298.7007 | -2.3200 | -2.3641 |
425
+ | 0.0077 | 2.89 | 5600 | 0.7520 | -4.5586 | -8.3485 | 0.7969 | 3.7899 | -340.4545 | -299.8206 | -2.3078 | -2.3517 |
426
+ | 0.0094 | 2.94 | 5700 | 0.7527 | -4.5542 | -8.3509 | 0.7812 | 3.7967 | -340.4790 | -299.7773 | -2.3062 | -2.3510 |
427
+ | 0.0054 | 2.99 | 5800 | 0.7520 | -4.5169 | -8.3079 | 0.7812 | 3.7911 | -340.0493 | -299.4038 | -2.3081 | -2.3530 |
428
+
429
+
430
+ ### Framework versions
431
+
432
+ - Transformers 4.35.0.dev0
433
+ - Pytorch 2.0.1+cu118
434
+ - Datasets 2.12.0
435
+ - Tokenizers 0.14.0
436
+
437
+ ## Citation
438
+
439
+ If you find Zephyr-7B-β is useful in your work, please cite it with:
440
+
441
+ ```
442
+ @misc{tunstall2023zephyr,
443
+ title={Zephyr: Direct Distillation of LM Alignment},
444
+ author={Lewis Tunstall and Edward Beeching and Nathan Lambert and Nazneen Rajani and Kashif Rasul and Younes Belkada and Shengyi Huang and Leandro von Werra and Clémentine Fourrier and Nathan Habib and Nathan Sarrazin and Omar Sanseviero and Alexander M. Rush and Thomas Wolf},
445
+ year={2023},
446
+ eprint={2310.16944},
447
+ archivePrefix={arXiv},
448
+ primaryClass={cs.LG}
449
+ }
450
+ ```
451
+ # [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
452
+ Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_HuggingFaceH4__zephyr-7b-beta)
453
+
454
+ | Metric | Value |
455
+ |-----------------------|---------------------------|
456
+ | Avg. | 52.15 |
457
+ | ARC (25-shot) | 62.03 |
458
+ | HellaSwag (10-shot) | 84.36 |
459
+ | MMLU (5-shot) | 61.07 |
460
+ | TruthfulQA (0-shot) | 57.45 |
461
+ | Winogrande (5-shot) | 77.74 |
462
+ | GSM8K (5-shot) | 12.74 |
463
+ | DROP (3-shot) | 9.66 |
464
+