README.md

---
library_name: transformers
license: apache-2.0
base_model: Qwen/Qwen2.5-1.5B-Instruct
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: Qwen2.5-1.5B-Instruct-SFT-BigmathV_Simple_Balanced-LR1.0e-5-EPOCHS2
  results: []
---

[**TinyV**]((https://arxiv.org/abs/2505.14625)) is a reward system for efficient RL post-training that detects false negatives in current rule-based verifiers and provides more accurate reward signals via a small LLM during RL training. Experiments show that TinyV incurs only 6% additional computational cost while significantly increasing both RL efficiency and final model performance.

- 📄 [Technical Report](https://arxiv.org/abs/2505.14625) - Including false negative analysis and theotical insights behind TinyV
- 💾 [Github Repo](https://github.com/uw-nsl/TinyV) - Access the complete pipeline for more efficient RL training via TinyV
- 🤗 [HF Collection](https://huggingface.co/collections/zhangchenxu/tinyv-682d5840c7e309217df625df) - Training Data, Benchmarks, and Model Artifact

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B-Instruct on [zhangchenxu/TinyV_Training_Data_Balanced](https://huggingface.co/datasets/zhangchenxu/TinyV_Training_Data_Balanced) dataset.

### Overview
![TinyV Pipeline](fn_tinyv_combine.png)

### How to use it?

Please refer to the codebase: [https://github.com/uw-nsl/TinyV](https://github.com/uw-nsl/TinyV) for details.

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 512
- total_eval_batch_size: 64
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 2.0

### Framework versions

- Transformers 4.48.3
- Pytorch 2.5.0
- Datasets 3.2.0
- Tokenizers 0.21.0