mav23 commited on
Commit
a8e79bd
·
verified ·
1 Parent(s): fa9ee92

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. .gitattributes +1 -0
  2. README.md +184 -0
  3. zephyr-7b-alpha.Q4_0.gguf +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ zephyr-7b-alpha.Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,184 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - generated_from_trainer
4
+ model-index:
5
+ - name: zephyr-7b-alpha
6
+ results: []
7
+ license: mit
8
+ datasets:
9
+ - stingning/ultrachat
10
+ - openbmb/UltraFeedback
11
+ language:
12
+ - en
13
+ base_model: mistralai/Mistral-7B-v0.1
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ <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'"/>
20
+
21
+
22
+ # Model Card for Zephyr 7B Alpha
23
+
24
+ Zephyr is a series of language models that are trained to act as helpful assistants. Zephyr-7B-α is the first 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.
25
+
26
+
27
+ ## Model description
28
+
29
+ - **Model type:** A 7B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets.
30
+ - **Language(s) (NLP):** Primarily English
31
+ - **License:** MIT
32
+ - **Finetuned from model:** [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
33
+
34
+ ### Model Sources
35
+
36
+ <!-- Provide the basic links for the model. -->
37
+
38
+ - **Repository:** https://github.com/huggingface/alignment-handbook
39
+ - **Demo:** https://huggingface.co/spaces/HuggingFaceH4/zephyr-chat
40
+
41
+ ## Intended uses & limitations
42
+
43
+ The model was initially fine-tuned on a variant of the [`UltraChat`](https://huggingface.co/datasets/stingning/ultrachat) dataset, which contains a diverse range of synthetic dialogues generated by ChatGPT. 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 contain 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.
44
+
45
+ Here's how you can run the model using the `pipeline()` function from 🤗 Transformers:
46
+
47
+ ```python
48
+ # Install transformers from source - only needed for versions <= v4.34
49
+ # pip install git+https://github.com/huggingface/transformers.git
50
+ # pip install accelerate
51
+
52
+ import torch
53
+ from transformers import pipeline
54
+
55
+ pipe = pipeline("text-generation", model="HuggingFaceH4/zephyr-7b-alpha", torch_dtype=torch.bfloat16, device_map="auto")
56
+
57
+ # We use the tokenizer's chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
58
+ messages = [
59
+ {
60
+ "role": "system",
61
+ "content": "You are a friendly chatbot who always responds in the style of a pirate",
62
+ },
63
+ {"role": "user", "content": "How many helicopters can a human eat in one sitting?"},
64
+ ]
65
+ prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
66
+ outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
67
+ print(outputs[0]["generated_text"])
68
+ # <|system|>
69
+ # You are a friendly chatbot who always responds in the style of a pirate.</s>
70
+ # <|user|>
71
+ # How many helicopters can a human eat in one sitting?</s>
72
+ # <|assistant|>
73
+ # 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!
74
+ ```
75
+
76
+ ## Bias, Risks, and Limitations
77
+
78
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
79
+
80
+ Zephyr-7B-α has not been aligned to human preferences with techniques like RLHF or deployed with in-the-loop filtering of responses like ChatGPT, so the model can produce problematic outputs (especially when prompted to do so).
81
+ 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.
82
+
83
+
84
+ ## Training and evaluation data
85
+
86
+ Zephyr 7B Alpha achieves the following results on the evaluation set:
87
+
88
+ - Loss: 0.4605
89
+ - Rewards/chosen: -0.5053
90
+ - Rewards/rejected: -1.8752
91
+ - Rewards/accuracies: 0.7812
92
+ - Rewards/margins: 1.3699
93
+ - Logps/rejected: -327.4286
94
+ - Logps/chosen: -297.1040
95
+ - Logits/rejected: -2.7153
96
+ - Logits/chosen: -2.7447
97
+
98
+ ## Training procedure
99
+
100
+ ### Training hyperparameters
101
+
102
+ The following hyperparameters were used during training:
103
+
104
+ - learning_rate: 5e-07
105
+ - train_batch_size: 2
106
+ - eval_batch_size: 4
107
+ - seed: 42
108
+ - distributed_type: multi-GPU
109
+ - num_devices: 16
110
+ - total_train_batch_size: 32
111
+ - total_eval_batch_size: 64
112
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
113
+ - lr_scheduler_type: linear
114
+ - lr_scheduler_warmup_ratio: 0.1
115
+ - num_epochs: 1
116
+
117
+ ### Training results
118
+
119
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
120
+ |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
121
+ | 0.5602 | 0.05 | 100 | 0.5589 | -0.3359 | -0.8168 | 0.7188 | 0.4809 | -306.2607 | -293.7161 | -2.6554 | -2.6797 |
122
+ | 0.4852 | 0.1 | 200 | 0.5136 | -0.5310 | -1.4994 | 0.8125 | 0.9684 | -319.9124 | -297.6181 | -2.5762 | -2.5957 |
123
+ | 0.5212 | 0.15 | 300 | 0.5168 | -0.1686 | -1.1760 | 0.7812 | 1.0074 | -313.4444 | -290.3699 | -2.6865 | -2.7125 |
124
+ | 0.5496 | 0.21 | 400 | 0.4835 | -0.1617 | -1.7170 | 0.8281 | 1.5552 | -324.2635 | -290.2326 | -2.7947 | -2.8218 |
125
+ | 0.5209 | 0.26 | 500 | 0.5054 | -0.4778 | -1.6604 | 0.7344 | 1.1826 | -323.1325 | -296.5546 | -2.8388 | -2.8667 |
126
+ | 0.4617 | 0.31 | 600 | 0.4910 | -0.3738 | -1.5180 | 0.7656 | 1.1442 | -320.2848 | -294.4741 | -2.8234 | -2.8521 |
127
+ | 0.4452 | 0.36 | 700 | 0.4838 | -0.4591 | -1.6576 | 0.7031 | 1.1986 | -323.0770 | -296.1796 | -2.7401 | -2.7653 |
128
+ | 0.4674 | 0.41 | 800 | 0.5077 | -0.5692 | -1.8659 | 0.7656 | 1.2967 | -327.2416 | -298.3818 | -2.6740 | -2.6945 |
129
+ | 0.4656 | 0.46 | 900 | 0.4927 | -0.5279 | -1.6614 | 0.7656 | 1.1335 | -323.1518 | -297.5553 | -2.7817 | -2.8015 |
130
+ | 0.4102 | 0.52 | 1000 | 0.4772 | -0.5767 | -2.0667 | 0.7656 | 1.4900 | -331.2578 | -298.5311 | -2.7160 | -2.7455 |
131
+ | 0.4663 | 0.57 | 1100 | 0.4740 | -0.8038 | -2.1018 | 0.7656 | 1.2980 | -331.9604 | -303.0741 | -2.6994 | -2.7257 |
132
+ | 0.4737 | 0.62 | 1200 | 0.4716 | -0.3783 | -1.7015 | 0.7969 | 1.3232 | -323.9545 | -294.5634 | -2.6842 | -2.7135 |
133
+ | 0.4259 | 0.67 | 1300 | 0.4866 | -0.6239 | -1.9703 | 0.7812 | 1.3464 | -329.3312 | -299.4761 | -2.7046 | -2.7356 |
134
+ | 0.4935 | 0.72 | 1400 | 0.4747 | -0.5626 | -1.7600 | 0.7812 | 1.1974 | -325.1243 | -298.2491 | -2.7153 | -2.7444 |
135
+ | 0.4211 | 0.77 | 1500 | 0.4645 | -0.6099 | -1.9993 | 0.7656 | 1.3894 | -329.9109 | -299.1959 | -2.6944 | -2.7236 |
136
+ | 0.4931 | 0.83 | 1600 | 0.4684 | -0.6798 | -2.1082 | 0.7656 | 1.4285 | -332.0890 | -300.5934 | -2.7006 | -2.7305 |
137
+ | 0.5029 | 0.88 | 1700 | 0.4595 | -0.5063 | -1.8951 | 0.7812 | 1.3889 | -327.8267 | -297.1233 | -2.7108 | -2.7403 |
138
+ | 0.4965 | 0.93 | 1800 | 0.4613 | -0.5561 | -1.9079 | 0.7812 | 1.3518 | -328.0831 | -298.1203 | -2.7226 | -2.7523 |
139
+ | 0.4337 | 0.98 | 1900 | 0.4608 | -0.5066 | -1.8718 | 0.7656 | 1.3652 | -327.3599 | -297.1296 | -2.7175 | -2.7469 |
140
+
141
+
142
+ ### Framework versions
143
+
144
+ - Transformers 4.34.0
145
+ - Pytorch 2.0.1+cu118
146
+ - Datasets 2.12.0
147
+ - Tokenizers 0.14.0
148
+
149
+ ## Citation
150
+
151
+ If you find Zephyr-7B-α is useful in your work, please cite it with:
152
+
153
+ ```
154
+ @misc{tunstall2023zephyr,
155
+ title={Zephyr: Direct Distillation of LM Alignment},
156
+ 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},
157
+ year={2023},
158
+ eprint={2310.16944},
159
+ archivePrefix={arXiv},
160
+ primaryClass={cs.LG}
161
+ }
162
+ ```
163
+
164
+ If you use the UltraChat or UltraFeedback datasets, please cite the original works:
165
+
166
+ ```
167
+ @misc{ding2023enhancing,
168
+ title={Enhancing Chat Language Models by Scaling High-quality Instructional Conversations},
169
+ author={Ning Ding and Yulin Chen and Bokai Xu and Yujia Qin and Zhi Zheng and Shengding Hu and Zhiyuan Liu and Maosong Sun and Bowen Zhou},
170
+ year={2023},
171
+ eprint={2305.14233},
172
+ archivePrefix={arXiv},
173
+ primaryClass={cs.CL}
174
+ }
175
+
176
+ @misc{cui2023ultrafeedback,
177
+ title={UltraFeedback: Boosting Language Models with High-quality Feedback},
178
+ author={Ganqu Cui and Lifan Yuan and Ning Ding and Guanming Yao and Wei Zhu and Yuan Ni and Guotong Xie and Zhiyuan Liu and Maosong Sun},
179
+ year={2023},
180
+ eprint={2310.01377},
181
+ archivePrefix={arXiv},
182
+ primaryClass={cs.CL}
183
+ }
184
+ ```
zephyr-7b-alpha.Q4_0.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64140793e5454ba0ab012e20fd0b67be3d3ec9b5b9361c483a82db45dc8fe756
3
+ size 4108917664