Skip to content

Instantly share code, notes, and snippets.

View ibndias's full-sized avatar
⚔️

Derry Pratama ibndias

⚔️
View GitHub Profile
@lewtun
lewtun / sft_trainer.py
Last active April 27, 2024 21:28
Fine-tuning Mistral 7B with TRL & DeepSpeed ZeRO-3
# This is a modified version of TRL's `SFTTrainer` example (https://github.com/huggingface/trl/blob/main/examples/scripts/sft_trainer.py),
# adapted to run with DeepSpeed ZeRO-3 and Mistral-7B-V1.0. The settings below were run on 1 node of 8 x A100 (80GB) GPUs.
#
# Usage:
# - Install the latest transformers & accelerate versions: `pip install -U transformers accelerate`
# - Install deepspeed: `pip install deepspeed==0.9.5`
# - Install TRL from main: pip install git+https://github.com/huggingface/trl.git
# - Clone the repo: git clone github.com/huggingface/trl.git
# - Copy this Gist into trl/examples/scripts
# - Run from root of trl repo with: accelerate launch --config_file=examples/accelerate_configs/deepspeed_zero3.yaml --gradient_accumulation_steps 8 examples/scripts/sft_trainer.py
@naufalso
naufalso / pdf-color-pages-classifier-and-counter.ipynb
Last active January 3, 2024 06:27
pdf-color-pages-classifier-and-counter.ipynb
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
@ibndias
ibndias / sft_trainer.py
Created February 8, 2024 11:51 — forked from lewtun/sft_trainer.py
Fine-tuning Mistral 7B with TRL & DeepSpeed ZeRO-3
# This is a modified version of TRL's `SFTTrainer` example (https://github.com/huggingface/trl/blob/main/examples/scripts/sft_trainer.py),
# adapted to run with DeepSpeed ZeRO-3 and Mistral-7B-V1.0. The settings below were run on 1 node of 8 x A100 (80GB) GPUs.
#
# Usage:
# - Install the latest transformers & accelerate versions: `pip install -U transformers accelerate`
# - Install deepspeed: `pip install deepspeed==0.9.5`
# - Install TRL from main: pip install git+https://github.com/huggingface/trl.git
# - Clone the repo: git clone github.com/huggingface/trl.git
# - Copy this Gist into trl/examples/scripts
# - Run from root of trl repo with: accelerate launch --config_file=examples/accelerate_configs/deepspeed_zero3.yaml --gradient_accumulation_steps 8 examples/scripts/sft_trainer.py