Downloads · 30 days
0
Mistral-7B-v0.1-onmt
OpenNMT/Mistral-7B-v0.1-onmt
Mistral-7B-v0.1-onmt is a machine learning model from OpenNMT. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
This model is the original Mistral-ai 7B v0.1 model converted to the OpenNMT-py format. By original, it means with interleaved rotary (option: rotaryinterleave=True)
Access
Public
Updated Nov 29, 2023
Repo size
14.5 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt14.5 GB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Nov 29, 2023
- Updated
- Nov 29, 2023
- SHA
- ab8b117d
Model card
From the Hugging Face model README
This model is the original Mistral-ai 7B v0.1 model converted to the OpenNMT-py format. By original, it means with interleaved rotary (option: rotary_interleave=True)
You need to install OpenNMT-py, instructions are here: https://github.com/OpenNMT/OpenNMT-py
Running inference:
Create a text input file with prompts (ex: "Show me some attractions in Boston")
then run:
onmt_translate --config mistral-inference.yaml --src input.txt --output output.txt
Running MMLU evaluation:
If you git clone the OpenNMT-py repo then you can run:
python eval_llm/MMLU/run_mmlu_opennmt.py --config mistral-inference.yaml
For this use case make sure you use max_length=1 in the config file
Finetuning:
Read this tuto: https://forum.opennmt.net/t/finetuning-llama-7b-13b-or-mosaicml-mpt-7b-reproduce-vicuna-alpaca/5272/56
onmt_train --config mistral-finetuning.yaml
- License
- apache-2.0
- Created
- Nov 29, 2023
- Updated
- Nov 29, 2023