Picture for Shuming Shi

Shuming Shi

Fuzzy Reasoning Chain (FRC): An Innovative Reasoning Framework from Fuzziness to Clarity

Add code
Sep 26, 2025
Viaarxiv icon

Quantitative Analysis of Performance Drop in DeepSeek Model Quantization

Add code
May 05, 2025
Viaarxiv icon

DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models

Add code
Mar 06, 2025
Viaarxiv icon

CuDIP: Enhancing Theorem Proving in LLMs via Curriculum Learning-based Direct Preference Optimization

Add code
Feb 25, 2025
Viaarxiv icon

Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability

Add code
Oct 15, 2024
Figure 1 for Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability
Figure 2 for Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability
Figure 3 for Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability
Figure 4 for Selection-p: Self-Supervised Task-Agnostic Prompt Compression for Faithfulness and Transferability
Viaarxiv icon

An Energy-based Model for Word-level AutoCompletion in Computer-aided Translation

Add code
Jul 29, 2024
Viaarxiv icon

Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning

Add code
Jun 25, 2024
Figure 1 for Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
Figure 2 for Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
Figure 3 for Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
Figure 4 for Not All Preference Pairs Are Created Equal: A Recipe for Annotation-Efficient Iterative Preference Learning
Viaarxiv icon

On the Transformations across Reward Model, Parameter Update, and In-Context Prompt

Add code
Jun 24, 2024
Viaarxiv icon

Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction

Add code
May 22, 2024
Figure 1 for Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction
Figure 2 for Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction
Figure 3 for Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction
Figure 4 for Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction
Viaarxiv icon

Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text

Add code
May 21, 2024
Figure 1 for Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text
Figure 2 for Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text
Figure 3 for Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text
Figure 4 for Spotting AI's Touch: Identifying LLM-Paraphrased Spans in Text
Viaarxiv icon