FRPO/qwen3-1.7b-a3_onpolicy-k1-cNone-clip0.2-mb1-eta100-bs64x5-n2-s800

TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 15, 2026Architecture:Transformer Featherless Exclusive Cold

FRPO/qwen3-1.7b-a3_onpolicy-k1-cNone-clip0.2-mb1-eta100-bs64x5-n2-s800 is a 2 billion parameter language model based on the Qwen3-1.7B architecture, fine-tuned using Reinforcement Learning (RL) with the FRPO algorithm. This model is a checkpoint from the KL-in-LLM-RL experiments, specifically optimized for on-policy learning. It is designed for research and development in RL-tuned language models, offering insights into the effects of specific RL configurations.

Loading preview...

Model Overview

FRPO/qwen3-1.7b-a3_onpolicy-k1-cNone-clip0.2-mb1-eta100-bs64x5-n2-s800 is a 2 billion parameter language model derived from the Qwen/Qwen3-1.7B base model. It has undergone Reinforcement Learning (RL) fine-tuning using the FRPO (Fictitious Reinforcement Policy Optimization) algorithm, as part of the KL-in-LLM-RL experimental series.

Key Characteristics

  • Base Model: Qwen3-1.7B, a 2 billion parameter architecture.
  • Fine-tuning Method: Reinforcement Learning (RL) using the FRPO algorithm, specifically configured for on-policy learning.
  • Experimental Checkpoint: This model represents a specific checkpoint (global_step_800) from the KL-in-LLM-RL experiments, providing a snapshot of the model's state under a particular RL configuration.
  • Weights: Provided in fp32 safetensors format, preserving the exact state as saved by the trainer without further post-processing.
  • Configuration Transparency: The specific run configuration used for training is encoded directly within the repository name, allowing for detailed understanding of its experimental setup.

Intended Use Cases

This model is primarily intended for:

  • RL Research: Investigating the effects of FRPO and on-policy learning in the context of large language models.
  • Experimental Analysis: Studying the performance and characteristics of RL-tuned Qwen3-1.7B under specific training parameters.
  • Comparative Studies: Serving as a baseline or comparison point for other RL fine-tuning approaches on similar base models.

Finetunes
1 models