-
BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs
Paper • 2510.04721 • Published -
FormalMATH: Benchmarking Formal Mathematical Reasoning of Large Language Models
Paper • 2505.02735 • Published • 34 -
PolyMath: Evaluating Mathematical Reasoning in Multilingual Contexts
Paper • 2504.18428 • Published -
MathConstruct: Challenging LLM Reasoning with Constructive Proofs
Paper • 2502.10197 • Published
Shuo Xing
shuoxing
AI & ML interests
MLLMs, LLMs
Recent Activity
updated
a model
1 day ago
shuoxing/qwen2-5-7b-full-pretrain-control-tweet-1m-en-no-packing-new-sft-bs32
published
a model
1 day ago
shuoxing/qwen2-5-7b-full-pretrain-control-tweet-1m-en-no-packing-new-sft-bs32
updated
a model
1 day ago
shuoxing/qwen2-5-7b-full-pretrain-mix-high-tweet-1m-en-no-packing-new-sft-bs32
Organizations
Papers on the reasoning, rewarding, and understanding of the MLLMs and LLMs
-
Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning
Paper • 2505.24726 • Published • 275 -
SynthRL: Scaling Visual Reasoning with Verifiable Data Synthesis
Paper • 2506.02096 • Published • 52 -
OThink-R1: Intrinsic Fast/Slow Thinking Mode Switching for Over-Reasoning Mitigation
Paper • 2506.02397 • Published • 35 -
ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models
Paper • 2505.24864 • Published • 141
-
BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs
Paper • 2510.04721 • Published -
FormalMATH: Benchmarking Formal Mathematical Reasoning of Large Language Models
Paper • 2505.02735 • Published • 34 -
PolyMath: Evaluating Mathematical Reasoning in Multilingual Contexts
Paper • 2504.18428 • Published -
MathConstruct: Challenging LLM Reasoning with Constructive Proofs
Paper • 2502.10197 • Published
Papers on the reasoning, rewarding, and understanding of the MLLMs and LLMs
-
Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning
Paper • 2505.24726 • Published • 275 -
SynthRL: Scaling Visual Reasoning with Verifiable Data Synthesis
Paper • 2506.02096 • Published • 52 -
OThink-R1: Intrinsic Fast/Slow Thinking Mode Switching for Over-Reasoning Mitigation
Paper • 2506.02397 • Published • 35 -
ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models
Paper • 2505.24864 • Published • 141
models
153
shuoxing/qwen2-5-7b-full-pretrain-control-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
333k
•
Updated
•
45
shuoxing/qwen2-5-7b-full-pretrain-mix-high-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
333k
•
Updated
•
53
shuoxing/qwen3-4b-full-pretrain-control-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
196k
•
Updated
•
49
shuoxing/qwen2-5-7b-full-pretrain-mix-mid-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
333k
•
Updated
•
28
shuoxing/qwen3-4b-full-pretrain-mix-high-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
196k
•
Updated
•
47
shuoxing/qwen-0_5b-full-pretrain-control-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
0.5B
•
Updated
•
8
shuoxing/qwen3-4b-full-pretrain-mix-mid-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
196k
•
Updated
•
50
shuoxing/qwen2-5-7b-full-pretrain-mix-low-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
333k
•
Updated
•
38
shuoxing/qwen-0_5b-full-pretrain-mix-high-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
0.5B
•
Updated
•
23
shuoxing/qwen-0_5b-full-pretrain-mix-mid-tweet-1m-en-no-packing-new-sft-bs32
Text Generation
•
0.5B
•
Updated
•
29