Gauri-2704/Local-LLM
🛠️ Enable offline NLP workflows with Local-LLM, a lightweight Python library for secure environments using the BERT model, ensuring reproducibility and reliability.
What's novel
🛠️ Enable offline NLP workflows with Local-LLM, a lightweight Python library for secure environments using the BERT model, ensuring reproducibility and reliability.
Code Analysis
9 files read · 3 roundsA from-scratch PyTorch BERT implementation with WordPiece tokenization, TF1-to-PyTorch checkpoint conversion, and a text classification fine-tuning pipeline — all without HuggingFace.
Strengths
Genuinely implements BERT from scratch with correct attention, LayerNorm, GELU variants, and residual connections; the TF→Torch converter does real name mapping and weight transposition; test suite is comprehensive with edge cases for tokenization, attention masking, weight init, and finetune policies.
Weaknesses
README is completely misleading (describes a nonexistent ZIP download workflow and `python -m local_llm` command); tensorflow is a hard dependency despite only being needed for the optional converter; dead code (run_epoch stub, unused sqlalchemy import) and no __main__.py entry point.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
76
Contributors
1
Files
26
Active weeks
6
Repository
Language
Python
Stars
1
Forks
0
License
—