finetune the language model for task of natural language inference
-
Updated
Jun 11, 2023 - Python
finetune the language model for task of natural language inference
Welcome to the Neural Language Interface (NLI) Explain project! This repository is dedicated to exploring and explaining the decision-making process of BERT models in the context of Natural Language Inference (NLI) tasks. We employ Feature Interaction methods to shed light on why BERT makes specific predictions in NLI.
This is the DEMO Vietnamese SBERT project that I put my heart into most clearly.
Application of Transformers for NLI task with SNLI dataset
LLM uncertainty attribution overlays: token uncertainty, diff, contrastive likelihood, and NLI span localization in Gradio.
MedQA-NLI is a comprehensive medical reasoning dataset comprising 42,889 instances designed for training and evaluating models on natural language inference (NLI) tasks in biomedical domains.
A contradiction-aware RAG pipeline that surfaces source disagreement instead of averaging it away. Extracts atomic claims with byte-exact provenance, canonicalizes them into a claim graph, classifies supports/refutes relations, and synthesizes answers that report conflicts with evidence.
This repository is to understand Attention mechanism for the Classification task. The task used here for explanation is Recognizing Textual Entailment. It is a Natural Language Inference task.
A vibe-coded news summarizer that combines Seq2Seq generation, NLI-based consistency checking, NER-driven entity tracking, and SBERT semantic embeddings to deliver context-aware summaries.
Hallucination-resistant multi-hop QA using hybrid BM25+FAISS retrieval, cross-encoder reranking, citation selection, and NLI-based verification. Evaluated on HotpotQA (7,405 examples), zero-shot.
📃⚖️🙅 Pipeline to detect contradictions in policy documents using NLP and transformers.
Learning to Infer from Unlabeled Data: A Semi-supervised Learning Approach for Robust Natural Language Inference
Breaking and fixing the SNLI benchmark. Generating challenge sets to expose artifacts and mitigating them with targeted ELECTRA fine-tuning.
AI-powered climate statement fact-checker using LangGraph, Azure OpenAI, and DeBERTa NLI to detect contradictions against the IPCC AR6 report
Fully offline, GPU-accelerated RAG system for confidential board documents. Hybrid retrieval (BM25 + BGE-M3 dense + RRF + cross-encoder reranking), Qwen3-14B on vLLM, NLI hallucination guard with governance-aware verb checks, page-level citations, AES-256 ephemeral sessions. Zero network egress.
Natural Language Understanding - Natural Language Inference Coursework
Bachelor Thesis comparing two Relation-Inference Datasets
To associate your repository with the nli topic, visit your repo's landing page and select "manage topics."