You can deploy this model and start earning money today!

1337xyz1337xyz/
qwen3-8b-sft-crafter-improved11-long-12091343-ckpt12000

Name: qwen3-8b-sft-crafter-improved11-long-12091343-ckpt12000
Availability: InStock

text-generation-inference

endpoints_compatible

Views

No views yet

Qwen3-8B SFT on Crafter (improved11-long)

This repo contains the model weights from an intermediate HF Trainer checkpoint.

Base model: Qwen/Qwen3-8B
Dataset: 1337xyz1337xyz/crafter-gptoss120b-xml-improved11-long-11712830 (Crafter trajectories)
Training: SFT on action-only targets (<action>...</action>), job id 12091343
Checkpoint: checkpoint-12000

Notes

Only model weights + config are uploaded. Optimizer state is intentionally omitted.
To load:

python

from transformers import AutoModelForCausalLM, AutoTokenizer
m = AutoModelForCausalLM.from_pretrained("1337xyz1337xyz/qwen3-8b-sft-crafter-improved11-long-12091343-ckpt12000")
tok = AutoTokenizer.from_pretrained("Qwen/Qwen3-8B", trust_remote_code=True)