natesgituser/SmolFactory at 598357a7d8a087543477f0aac96b9910dc8dbbef

SmolFactory / scripts

291 kB

3 contributors

improves launcher with model family and defaults based on options, updates trl trainer , removes trl config paths by switching to trainingarguments class , tokenizer parameters updated to sfttrainer , resolves evaluation_strategy error

598357a 5 months ago

dataset_tonic
adds single token logic read/write , adds gpt-oss demo space , adds spaces refactor , adds new version of track tonic , adds logic in launch.sh 5 months ago
model_tonic
adds harmony format , configurable gpt-oss parameters, launch.sh logic , improved templates for legml gpt-oss training, dynamic results directory and improve model pushing 5 months ago
rescue
cleanup a bit the files 6 months ago
trackio_tonic
hide all tokens in logs, never persist to disk, remove max_seq_length from config, add to trainer 5 months ago
training
improves launcher with model family and defaults based on options, updates trl trainer , removes trl config paths by switching to trainingarguments class , tokenizer parameters updated to sfttrainer , resolves evaluation_strategy error 5 months ago
check_dependencies.py

2.09 kB

solves hf cli error 6 months ago
deploy_demo_space.py

25.2 kB

hide all tokens in logs, never persist to disk, remove max_seq_length from config, add to trainer 5 months ago
validate_hf_token.py

2.73 kB

adds new hf cli 6 months ago