Modelpublic

arvindcr4/tinker-rl-arch_gsm8k_gpt-oss-20b-gpt-oss-20b

sourceHugging Faceapache-2.0updated 5mo agoView on Hugging Face

0likes

Model Card

tinker-rl-archgsm8kgpt-oss-20b-gpt-oss-20b

LoRA adapters trained with GRPO on top of openai/gpt-oss-20b using the Tinker cloud training service. Part of the TinkerRL-Bench release for our NeurIPS submission "A Unified Benchmark for RL Post-Training of Language Models" (repo).

Training configuration


Base model	`openai/gpt-oss-20b`
Experiment tag	`arch_gsm8k_gpt-oss-20b`
Campaign	`None`
Task	`gsm8k`
Seed	`42`
LoRA rank	`32`
Learning rate	`3e-05`
Group size	`8`
Training steps	`30`
Platform	Tinker (`tinker`)
Training run ID	`c465d654-a00f-5821-9c11-0e40f0eecf9f`

Metrics

Metric	Value
Last-10 reward avg	0.875
Peak reward	1.0

Checkpoints in this repo

Step	Original Tinker URI	Local path
`sampler_weights/final`	tinker://c465d654-a00f-5821-9c11-0e40f0eecf9f:train:0/sampler_weights/final	`final`

How to load

python

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = "openai/gpt-oss-20b"
adapter = "arvindcr4/tinker-rl-arch_gsm8k_gpt-oss-20b-gpt-oss-20b"

tok = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(base, torch_dtype="auto", device_map="auto")
model = PeftModel.from_pretrained(model, adapter, subfolder="final")  # or "<step>"

Companion releases

—Dataset: `arvindcr4/tinker-rl-bench-wandb` — all 334 W&B runs + 9,255 history rows
—Manifest: `arvindcr4/tinker-rl-bench-checkpoints` — full catalogue of every Tinker URI
—Code: `pes-llm-research/tinker-rl-lab`

Citation

bibtex

@misc{tinkerrlbench2026,
  title   = {A Unified Benchmark for RL Post-Training of Language Models},
  author  = {Arvind, C. R. and Jeyaraj, Sandhya},
  year    = {2026},
  note    = {NeurIPS submission, https://github.com/pes-llm-research/tinker-rl-lab}
}

License

Apache 2.0. The underlying base model retains its original license — please check openai/gpt-oss-20b for any usage restrictions.