Mistral-OpenOrca

OpenOrca - Mistral - 7B - 8k

We have used our own OpenOrca dataset to fine-tune on top of Mistral 7B. This dataset is our attempt to reproduce the dataset generated for Microsoft Research's Orca Paper. We use OpenChat packing, trained with Axolotl.

This release is trained on a curated filtered subset of most of our GPT-4 augmented data. It is the same subset of our data as was used in our OpenOrcaxOpenChat-Preview2-13B model.

HF Leaderboard evals place this model as #1 for all models smaller than 30B at release time, outperforming all other 7B and 13B models!

This release provides a first: a fully open model with class-breaking performance, capable of running fully accelerated on even moderate consumer GPUs. Our thanks to the Mistral team for leading the way here.

We affectionately codename this model: "MistralOrca"

If you'd like to try the model now, we have it running on fast GPUs unquantized: https://huggingface.co/spaces/Open-Orca/Mistral-7B-OpenOrca

Want to visualize our full (pre-filtering) dataset? Check out our Nomic Atlas Map.

We are in-process with training more models, so keep a look out on our org for releases coming soon with exciting partners.

We will also give sneak-peak announcements on our Discord, which you can find here:

https://AlignmentLab.ai

or check the OpenAccess AI Collective Discord for more information about Axolotl trainer here:

https://discord.gg/5y8STgB3P3

Quantized Models

Quantized versions of this model are generously made available by TheBloke.

• AWQ: https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-AWQ
• GPTQ: https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-GPTQ
• GGUF: https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-GGUF

Prompt Template

We used OpenAI's Chat Markup Language (ChatML) format, with and tokens added to support this.

This means that, e.g., in oobabooga the "MPT-Chat" instruction template should work, as it also uses ChatML.

This formatting is also available via a pre-defined Transformers chat template, which means that lists of messages can be formatted for you with the apply_chat_template() method:

chat = [
  {"role": "system", "content": "You are MistralOrca, a large language model trained by Alignment Lab AI. Write out your reasoning step-by-step to be sure you get the right answers!"}
  {"role": "user", "content": "How are you?"},
  {"role": "assistant", "content": "I am doing well!"},
  {"role": "user", "content": "Please tell me about how mistral winds have attracted super-orcas."},
]
tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)

which will yield:

You are MistralOrca, a large language model trained by Alignment Lab AI. Write out your reasoning step-by-step to be sure you get the right answers!
How are you?
I am doing well!
Please tell me about how mistral winds have attracted super-orcas.

If you use tokenize=True and return_tensors="pt" instead, then you will get a tokenized and formatted conversation ready to pass to model.generate().

Inference

Note that you need the development snapshot of Transformers currently, as support for Mistral hasn't been released into PyPI yet:

pip install git+https://github.com/huggingface/transformers

Evaluation

HuggingFace Leaderboard Performance

We have evaluated using the methodology and tools for the HuggingFace Leaderboard, and find that we have dramatically improved upon the base model. We find 106% of the base model's performance on HF Leaderboard evals, averaging 65.84.

At release time, this beats all 7B and 13B models!

This is also 98.6% of Llama2-70b-chat's performance!

We use Language Model Evaluation Harness to run the benchmark tests above, using the same version as the HuggingFace LLM Leaderboard.

AGIEval Performance

We compare our results to the base Mistral-7B model (using LM Evaluation Harness).

We find 129% of the base model's performance on AGI Eval, averaging 0.397. As well, we significantly improve upon the official mistralai/Mistral-7B-Instruct-v0.1 finetuning, achieving 119% of their performance.

BigBench-Hard Performance

We find 119% of the base model's performance on BigBench-Hard, averaging 0.416.

GPT4ALL Leaderboard Performance

We gain a slight edge over our previous releases, again topping the leaderboard, averaging 72.38.

MT-Bench Performance

MT-Bench uses GPT-4 as a judge of model response quality, across a wide range of challenges. We find our performance is on-par with Llama2-70b-chat, averaging 6.86.

Dataset

We used a curated, filtered selection of most of the GPT-4 augmented data from our OpenOrca dataset, which aims to reproduce the Orca Research Paper dataset.

Training

We trained with 8x A6000 GPUs for 62 hours, completing 4 epochs of full fine tuning on our dataset in one training run. Commodity cost was ~$400