agurung/flawed-fictions-gemma-3-4b-lengthpenalty

Flawed Fictions GRPO (gemma3_4b_grpo_lengthpenalty)

Training Details

Base model	google/gemma-3-4b-it
Task	Continuity error detection (\boxed{Yes} / \boxed{No})
W&B project	flawed_fictions_rl
W&B group	grpo_flawed_fictions_gemma3_4b_lengthpenalty
W&B runs	dva1to9i
Training script	scripts/gemma3_4b_grpo_lengthpenalty.sh

Checkpoint Revisions

• Branch head (latest): main
• Per-checkpoint tags: main-step-<N>

Usage

from transformers import AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained(
    "agurung/flawed-fictions-gemma-3-4b-lengthpenalty",
    revision="main",
    device_map="auto",
    torch_dtype="auto",
)