Edit model card
Zephyr 7B Gemma Logo

Model Card for Zephyr 7B Gemma

Zephyr is a series of language models that are trained to act as helpful assistants. Zephyr 7B Gemma is the third model in the series, and is a fine-tuned version of google/gemma-7b that was trained on on a mix of publicly available, synthetic datasets using Direct Preference Optimization (DPO). You can reproduce the training of this model via the recipe provided in the Alignment Handbook.

Model description

  • Model type: A 7B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets.
  • Language(s) (NLP): Primarily English
  • License: Gemma Terms of Use
  • Finetuned from model: google/gemma-7b

Model Sources

Performance

Model MT Bench⬇️ IFEval
zephyr-7b-gemma-v0.1 7.81 28.76
zephyr-7b-beta 7.34 43.81
google/gemma-7b-it 6.38 38.01
Model AGIEval GPT4All TruthfulQA BigBench Average ⬇️
zephyr-7b-beta 37.52 71.77 55.26 39.77 51.08
zephyr-7b-gemma-v0.1 34.22 66.37 52.19 37.10 47.47
mlabonne/Gemmalpaca-7B 21.6 40.87 44.85 30.49 34.45
google/gemma-7b-it 21.33 40.84 41.70 30.25 33.53
Details of AGIEval, GPT4All, TruthfulQA, BigBench

AGIEval

Task Version Metric Value Stderr
agieval_aqua_rat 0 acc 21.65 Β± 2.59
acc_norm 25.20 Β± 2.73
agieval_logiqa_en 0 acc 34.72 Β± 1.87
acc_norm 35.94 Β± 1.88
agieval_lsat_ar 0 acc 19.57 Β± 2.62
acc_norm 21.74 Β± 2.73
agieval_lsat_lr 0 acc 30.59 Β± 2.04
acc_norm 32.55 Β± 2.08
agieval_lsat_rc 0 acc 49.07 Β± 3.05
acc_norm 42.75 Β± 3.02
agieval_sat_en 0 acc 54.85 Β± 3.48
acc_norm 53.40 Β± 3.48
agieval_sat_en_without_passage 0 acc 37.38 Β± 3.38
acc_norm 33.98 Β± 3.31
agieval_sat_math 0 acc 30.91 Β± 3.12
acc_norm 28.18 Β± 3.04

Average: 34.22%

GPT4All

Task Version Metric Value Stderr
arc_challenge 0 acc 49.15 Β± 1.46
acc_norm 52.47 Β± 1.46
arc_easy 0 acc 77.44 Β± 0.86
acc_norm 74.75 Β± 0.89
boolq 1 acc 79.69 Β± 0.70
hellaswag 0 acc 60.59 Β± 0.49
acc_norm 78.00 Β± 0.41
openbookqa 0 acc 29.20 Β± 2.04
acc_norm 37.80 Β± 2.17
piqa 0 acc 76.82 Β± 0.98
acc_norm 77.80 Β± 0.97
winogrande 0 acc 64.09 Β± 1.35

Average: 66.37%

TruthfulQA

Task Version Metric Value Stderr
truthfulqa_mc 1 mc1 35.74 Β± 1.68
mc2 52.19 Β± 1.59

Average: 52.19%

Bigbench

Task Version Metric Value Stderr
bigbench_causal_judgement 0 multiple_choice_grade 53.68 Β± 3.63
bigbench_date_understanding 0 multiple_choice_grade 59.89 Β± 2.55
bigbench_disambiguation_qa 0 multiple_choice_grade 30.23 Β± 2.86
bigbench_geometric_shapes 0 multiple_choice_grade 11.42 Β± 1.68
exact_str_match 0.00 Β± 0.00
bigbench_logical_deduction_five_objects 0 multiple_choice_grade 28.40 Β± 2.02
bigbench_logical_deduction_seven_objects 0 multiple_choice_grade 19.14 Β± 1.49
bigbench_logical_deduction_three_objects 0 multiple_choice_grade 44.67 Β± 2.88
bigbench_movie_recommendation 0 multiple_choice_grade 26.80 Β± 1.98
bigbench_navigate 0 multiple_choice_grade 50.00 Β± 1.58
bigbench_reasoning_about_colored_objects 0 multiple_choice_grade 52.75 Β± 1.12
bigbench_ruin_names 0 multiple_choice_grade 33.04 Β± 2.22
bigbench_salient_translation_error_detection 0 multiple_choice_grade 33.37 Β± 1.49
bigbench_snarks 0 multiple_choice_grade 48.62 Β± 3.73
bigbench_sports_understanding 0 multiple_choice_grade 58.11 Β± 1.57
bigbench_temporal_sequences 0 multiple_choice_grade 37.20 Β± 1.53
bigbench_tracking_shuffled_objects_five_objects 0 multiple_choice_grade 20.08 Β± 1.13
bigbench_tracking_shuffled_objects_seven_objects 0 multiple_choice_grade 15.77 Β± 0.87
bigbench_tracking_shuffled_objects_three_objects 0 multiple_choice_grade 44.67 Β± 2.88

Average: 37.1%

Intended uses & limitations

The model was initially fine-tuned on the DEITA 10K dataset, which contains a diverse range of synthetic dialogues generated by ChatGPT. We then further aligned the model with πŸ€— TRL's DPOTrainer on the argilla/dpo-mix-7k dataset, which contains 7k 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 to test its capabilities.

Here's how you can run the model using the pipeline() function from πŸ€— Transformers:

# pip install transformers>=4.38.2
# pip install accelerate

import torch
from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="HuggingFaceH4/zephyr-7b-gemma-v0.1",
    device_map="auto",
    torch_dtype=torch.bfloat16,
)
messages = [
    {
        "role": "system",
        "content": "",  # Model not yet trained for follow this
    },
    {"role": "user", "content": "How many helicopters can a human eat in one sitting?"},
]
outputs = pipe(
    messages,
    max_new_tokens=128,
    do_sample=True,
    temperature=0.7,
    top_k=50,
    top_p=0.95,
    stop_sequence="<|im_end|>",
)
print(outputs[0]["generated_text"][-1]["content"])
# It is not possible for a human to eat a helicopter in one sitting, as a
# helicopter is a large and inedible machine. Helicopters are made of metal,
# plastic, and other materials that are not meant to be consumed by humans.
# Eating a helicopter would be extremely dangerous and would likely cause
# serious health problems, including choking, suffocation, and poisoning. It is
# important to only eat food that is safe and intended for human consumption.

Bias, Risks, and Limitations

Zephyr 7B Gemma 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). It is also unknown what the size and composition of the corpus was used to train the base model (google/gemma-7b), however it is likely to have included a mix of Web data and technical sources like books and code. See the StarCoder2 model card for an example of this.

Training and evaluation data

This model is a fine-tuned version of HuggingFaceH4/zephyr-7b-gemma-sft-v0.1 on the argilla/dpo-mix-7k dataset.

It achieves the following results on the evaluation set:

  • Loss: 0.4695
  • Rewards/chosen: -3.3746
  • Rewards/rejected: -4.9715
  • Rewards/accuracies: 0.7188
  • Rewards/margins: 1.5970
  • Logps/rejected: -459.4853
  • Logps/chosen: -429.9115
  • Logits/rejected: 86.4684
  • Logits/chosen: 92.8200

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-07
  • train_batch_size: 2
  • eval_batch_size: 4
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 8
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 128
  • total_eval_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Rewards/chosen Rewards/rejected Rewards/accuracies Rewards/margins Logps/rejected Logps/chosen Logits/rejected Logits/chosen
0.1923 1.9 100 0.4736 -3.4575 -4.9556 0.75 1.4980 -459.1662 -431.5707 86.3863 92.7360

Framework versions

  • Transformers 4.39.0.dev0
  • Pytorch 2.1.2+cu121
  • Datasets 2.14.6
  • Tokenizers 0.15.1

Citation Information

If you find this model useful in your work, please consider citing the Zephyr technical report:

@misc{tunstall2023zephyr,
      title={Zephyr: Direct Distillation of LM Alignment}, 
      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 ClΓ©mentine Fourrier and Nathan Habib and Nathan Sarrazin and Omar Sanseviero and Alexander M. Rush and Thomas Wolf},
      year={2023},
      eprint={2310.16944},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

You may also wish to cite the creators of this model as well:

@misc{zephyr_7b_gemma,
  author = {Lewis Tunstall and Philipp Schmid},
  title = {Zephyr 7B Gemma},
  year = {2024},
  publisher = {Hugging Face},
  journal = {Hugging Face repository},
  howpublished = {\url{https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-v0.1}}
}

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 62.41
AI2 Reasoning Challenge (25-Shot) 58.45
HellaSwag (10-Shot) 83.48
MMLU (5-Shot) 60.68
TruthfulQA (0-shot) 52.07
Winogrande (5-shot) 74.19
GSM8k (5-shot) 45.56
Downloads last month
928
Safetensors
Model size
8.54B params
Tensor type
BF16
Β·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for HuggingFaceH4/zephyr-7b-gemma-v0.1

Base model

google/gemma-7b
Finetuned
(32)
this model
Adapters
4 models
Quantizations
1 model

Dataset used to train HuggingFaceH4/zephyr-7b-gemma-v0.1

Spaces using HuggingFaceH4/zephyr-7b-gemma-v0.1 99

Collection including HuggingFaceH4/zephyr-7b-gemma-v0.1

Evaluation results