---
base_model: Qwen/Qwen3-8B
license: apache-2.0
library_name: transformers
pipeline_tag: text-generation
language: [en]
tags:
  - qwen3
  - meta-rl
  - in-context-learning
  - grpo
---

# LatentGym_Qwen3-8B_10episodes_SingleLatent_number_guessing

GRPO-fine-tuned Qwen3-8B for single-latent training on number_guessing (N=10). Part of the [LatentGym](https://huggingface.co/collections/namkoong-lab/latentgym) testbed.

## Environments & training latents

| Env | Latents seen during training |
|---|---|
| number_guessing | `set_of_3` |

## Training hyperparameters

| | |
|---|---|
| Base model | `Qwen/Qwen3-8B` |
| Algorithm | GRPO |
| Optimizer | AdamW (β₁=0.9, β₂=0.999) |
| Learning rate | 5e-07 |
| LR schedule | constant_with_warmup |
| Weight decay | 0.01 |
| Max grad norm | 1.0 |
| KL coefficient β | 0.04 |
| Clip range ε | 0.2 |
| Train batch (prompts) | 32 |
| Mini-batch | 1 |
| Rollouts per prompt | 8 |
| Episodes per trajectory (N) | 10 |
| Reward Φ | Σᵢ Gᵢ (cumulative) |
| Epochs | 15 |
| Max generation length | 64 |
| Sampling (train) | T=0.8, top-p=0.95 |
| Seed | 263 |