Papers with full-model fine-tuning
InteMATs: Integrating Granularity-Specific Multilingual Adapters for Cross-Lingual Transfer (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing work relies on full-model fine-tuning on large parallel datasets to enhance cross-lingual alignment of MLLMs. |
| Approach: | They propose an approach that integrates multilingual adapters trained on texts of different levels of granularity into multilingual models. |
| Outcome: | The proposed approach improves the performance of multilingual language models on low-resource languages. |
GRASPrune: Global Gating for Budgeted Structured Pruning of Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | Large language models are expensive to serve because dense FFN blocks, multi-head attention, and KV caches dominate memory. |
| Approach: | They propose a global budgeted structured pruning framework that prunes FFN channels and attention KV head groups under a single global parameter budget. |
| Outcome: | The proposed model removes 50% of parameters and achieves 12.18 perplexity on WikiText-2 while maintaining competitive average zero-shot accuracy on five downstream benchmarks. |
PPT: Pre-trained Prompt Tuning for Few-shot Learning (2022.acl-long)
Copied to clipboard
| Challenge: | Prompt tuning for pre-trained language models has shown remarkable performance . however, prompt tuning is still not fully explored . |
| Approach: | They propose to pre-train prompts by adding soft prompts into the pre-training stage to obtain a better initialization. |
| Outcome: | The proposed framework outperforms full-model tuning under full-data and few-shot learning settings. |
SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs (2025.acl-long)
Copied to clipboard
| Challenge: | Existing approaches to continuous-space reasoning focus on hard token decoding and suffer from catastrophic forgetting. |
| Approach: | They propose a method that generates instance-specific soft thought tokens as the initial chain of thoughts and maps them into the LLM’s representation space via a trainable projection module. |
| Outcome: | The proposed method improves LLM reasoning performance through supervised, parameter-efficient fine-tuning. |