Tools - a trl-lib Collection

trl-lib 's Collections

Papers implemented in TRL

Prompt datasets (for GRPO, RLOO, ...)

Preference datasets (for DPO, reward modeling)

Unpaired preference datasets (KTO)

Models from the docs

Comparing DPO with IPO and KTO

Tools

updated 18 minutes ago

Spaces that help you train with TRL.

Running

Agents

4

Trackio

🚀

4

Track and visualize data streams in real-time

Note Training curves from the TRL docs, logged with Trackio
Running

Agents

6

Recommend vLLM Memory

😻

6

Estimate GPU memory usage for model training

Note Pick vllm_gpu_memory_utilization for GRPO and RLOO
Sleeping

Agents

Chat Template Inspector

📚

Inspect and test chat templates for Hugging Face models

Note Inspect a model's chat template
Paused

Agents

8

Dataset Length Profiler

👁

8

Estimate optimal max_length for SFT training with token analysis

Note Profile token lengths to choose max_length