Instructions to use Shawnno/RL-sft-adapter with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Shawnno/RL-sft-adapter with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-3B-Instruct") model = PeftModel.from_pretrained(base_model, "Shawnno/RL-sft-adapter") - Notebooks
- Google Colab
- Kaggle
Commit History
-
v3 SFT: lr=2e-4 NEFTune=0 no-oversample | eval_loss 1.039->0.992 train_loss 0.941 acc 81.1%
Verified signature11fc04f -
SFT v2: Qwen2.5-3B QLoRA, eval_loss=0.983, acc=73.9%
Verified signature8fabe65
-
remove train.log
Verified signature230e3e9 -
remove checkpoint-600/vocab.json
Verified signaturec2324be -
remove checkpoint-600/training_args.bin
Verified signature90bdb18 -
remove checkpoint-600/trainer_state.json
Verified signaturefa807bb -
remove checkpoint-600/tokenizer_config.json
Verified signature287e788 -
remove checkpoint-600/tokenizer.json
Verified signaturebcd33e7 -
remove checkpoint-600/special_tokens_map.json
Verified signature12028bd -
remove checkpoint-600/scheduler.pt
Verified signature83e884f -
remove checkpoint-600/rng_state.pth
Verified signature86de83e -
remove checkpoint-600/optimizer.pt
Verified signature0b2c920 -
remove checkpoint-600/merges.txt
Verified signatured8fb54e -
remove checkpoint-600/added_tokens.json
Verified signature8633c75 -
remove checkpoint-600/adapter_model.safetensors
Verified signature92d74d1 -
remove checkpoint-600/adapter_config.json
Verified signature805f85b -
remove checkpoint-600/README.md
Verified signature4443435 -
remove checkpoint-1104/vocab.json
Verified signature96663c8 -
remove checkpoint-1104/training_args.bin
Verified signature50a4262 -
remove checkpoint-1104/trainer_state.json
Verified signaturec553aee -
remove checkpoint-1104/tokenizer_config.json
Verified signature885a2df -
remove checkpoint-1104/tokenizer.json
Verified signature85f9c36 -
remove checkpoint-1104/special_tokens_map.json
Verified signature16d5717 -
remove checkpoint-1104/scheduler.pt
Verified signature8f2c756 -
remove checkpoint-1104/rng_state.pth
Verified signature261622c -
remove checkpoint-1104/optimizer.pt
Verified signature20c4705 -
remove checkpoint-1104/merges.txt
Verified signature4b00a28 -
remove checkpoint-1104/added_tokens.json
Verified signature8f8cd13 -
remove checkpoint-1104/adapter_model.safetensors
Verified signatureebb6f6a -
remove checkpoint-1104/adapter_config.json
Verified signatureec7f960 -
remove checkpoint-1104/README.md
Verified signature3ab5d6b -
v2: sft quality improvements - loss masking, neftune, system prompt, oversampling
Verified signature75af566
-
Upload folder using huggingface_hub
Verified signatureb0e8d85 -
initial commit
Verified signature2a95b69