Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# Prompt-R1: Enhancing LLM interaction on behalf of humans
|
| 2 |
|
| 3 |
<div align="center">
|
|
@@ -19,7 +37,7 @@
|
|
| 19 |
## Overview
|
| 20 |
|
| 21 |
<div align="center">
|
| 22 |
-
<img src="
|
| 23 |
</div>
|
| 24 |
|
| 25 |
**Prompt-R1** has addressed a critical challenge in interacting with large language models (LLMs)—the inability of users to provide accurate and effective interaction prompts for complex tasks. **Prompt-R1** is an **end-to-end reinforcement learning (RL)** framework that enhances the performance of LLMs by facilitating **collaborative automatic prompting** between a small-scale LLM and a large-scale LLM. **Prompt-R1**, through **multi-turn prompt interaction**, significantly improves the generation quality and reasoning accuracy of large-scale LLMs, enabling better task-solving performance without requiring user expertise in prompt formulation.
|
|
@@ -27,7 +45,7 @@
|
|
| 27 |
|
| 28 |
|
| 29 |
<div align="center">
|
| 30 |
-
<img src="
|
| 31 |
</div>
|
| 32 |
|
| 33 |
By integrating **collaborative prompting** and **reinforcement learning**, **Prompt-R1** offers a **plug-and-play framework** that supports both **inference** and **training** with **various large-scale LLMs** as the environment.
|
|
@@ -35,7 +53,7 @@ By integrating **collaborative prompting** and **reinforcement learning**, **Pro
|
|
| 35 |
## Experimental Results
|
| 36 |
**Results of Different Large language models:**
|
| 37 |
<div align="center">
|
| 38 |
-
<img src="
|
| 39 |
</div>
|
| 40 |
|
| 41 |
|
|
|
|
| 1 |
+
---
|
| 2 |
+
# YAML 元数据块 (Model Card Header)
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
license: apache-2.0
|
| 6 |
+
model_name: Prompt-R1 Model
|
| 7 |
+
tags:
|
| 8 |
+
- text-generation
|
| 9 |
+
- reinforcement-learning
|
| 10 |
+
- nlp
|
| 11 |
+
- transformers
|
| 12 |
+
- safetensors
|
| 13 |
+
# 关联的数据集
|
| 14 |
+
datasets:
|
| 15 |
+
- QwenQKing/Prompt-R1
|
| 16 |
+
# 决定网页右侧的推理小组件类型
|
| 17 |
+
pipeline_tag: text-generation
|
| 18 |
+
---
|
| 19 |
# Prompt-R1: Enhancing LLM interaction on behalf of humans
|
| 20 |
|
| 21 |
<div align="center">
|
|
|
|
| 37 |
## Overview
|
| 38 |
|
| 39 |
<div align="center">
|
| 40 |
+
<img src="image/2-QA.png" width="80%"/>
|
| 41 |
</div>
|
| 42 |
|
| 43 |
**Prompt-R1** has addressed a critical challenge in interacting with large language models (LLMs)—the inability of users to provide accurate and effective interaction prompts for complex tasks. **Prompt-R1** is an **end-to-end reinforcement learning (RL)** framework that enhances the performance of LLMs by facilitating **collaborative automatic prompting** between a small-scale LLM and a large-scale LLM. **Prompt-R1**, through **multi-turn prompt interaction**, significantly improves the generation quality and reasoning accuracy of large-scale LLMs, enabling better task-solving performance without requiring user expertise in prompt formulation.
|
|
|
|
| 45 |
|
| 46 |
|
| 47 |
<div align="center">
|
| 48 |
+
<img src="image/1-overview.png" width="90%"/>
|
| 49 |
</div>
|
| 50 |
|
| 51 |
By integrating **collaborative prompting** and **reinforcement learning**, **Prompt-R1** offers a **plug-and-play framework** that supports both **inference** and **training** with **various large-scale LLMs** as the environment.
|
|
|
|
| 53 |
## Experimental Results
|
| 54 |
**Results of Different Large language models:**
|
| 55 |
<div align="center">
|
| 56 |
+
<img src="image/6-radar.png" width="100%"/>
|
| 57 |
</div>
|
| 58 |
|
| 59 |
|