FairPut - Machine Learning Fairness Framework with LightGBM — Explainability, Robustness, Fairness (by @firmai)
Context-aware Biases for Length Extrapolation
GPT* - Training faster small transformers using ALiBi, Parallel Residual Connections and more!
Taxonomy Completion with Embedding Quantization and an LLM-based Pipeline: A Case Study in Computational Linguistics
CARE: Coherent Actionable Recourse based on Sound Counterfactual Explanations
CPU-first long-context KV memory optimization with deterministic page regeneration and exact folded attention.
Predicting loan defaults using Random Forest with SMOTE, SHAP, LIME, and fairness evaluation via AIF360 and Fairlearn
Testing MLServer and Alibi
Transformer Models for Humorous Text Generation. Fine-tuned on Russian jokes dataset with ALiBi, RoPE, GQA, and SwiGLU.Plus a custom Byte-level BPE tokenizer.
PyTorch implementation of Transformer encoder and GPT-style decoder with language modeling, speech classification, and ALiBi positional encoding experiments.
Report about "Digital Alibi" for academic course of Digital Forensics
From-scratch PyTorch Transformer encoders and a causal decoder with attention variants, reproducible NLP benchmarks, tests, visualizations, and Apple MPS support.
An XGBoost model in Python that classifies if a customer will cancel his/her hotel booking or not. I also use counterfactuals guided by prototypes from the Alibi package to explore the minimum changes needed to flip a prediction from canceled to not canceled and vice versa.
Comparison of positional encoding schemes in transformer
Counterfactual Explanations, a project for CITS4404 - Artificial Intelligence and Adaptive Systems.
RoPE, ALiBi, sinusoidal, and learned positional encodings — compared experimentally
ThreatLens v2 is a cutting-edge project that combines data analysis and cybersecurity to detect and analyze abnormal access patterns in web logs
To associate your repository with the alibi topic, visit your repo's landing page and select "manage topics."