Skip to content
View cluebbers's full-sized avatar
3 followers · 0 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report cluebbers

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
cluebbers/README.md

Header

Hello, I am Christopher L.

I fine-tune and red-team LLMs for safety and reliability. Recent: DPO pipeline (+3 pp paraphrase accuracy, +7 pp human preference), adversarial red-teaming (refusal rate 6% → 89%), METEOR reproducibility audit (10-point gap found). Currently: shutdown-resistance evaluations at Algoverse. Published on arXiv, 1st place Apart Research hackathon (35+ teams).

Pinned Loading

  1. ai-incident-forecasting ai-incident-forecasting Public

    🥇 1st place, Apart Research AI Forecasting Hackathon (35+ teams). Poisson GAM + multinomial share model forecasting AI incidents with calibrated 90% prediction intervals and rolling backtests. Pred…

    Jupyter Notebook 4

  2. dpo-rlhf-paraphrase-types dpo-rlhf-paraphrase-types Public

    End-to-end DPO fine-tuning pipeline for paraphrase-type generation (M.Sc. thesis, arXiv:2506.02018). DPO on 1,040 human-ranked pairs raised type accuracy +3 pp and human preference +7 pp over SFT b…

    Jupyter Notebook

  3. adverserial-paraphrasing adverserial-paraphrasing Public

    Red-teaming harness for open-weight LLMs (LLaMA, Mistral, Pythia). LoRA-SFT on 580 examples raised refusal rate from ~6% to 89% and cut harmful replies to 8%. Includes adversarial prompt dataset, S…

    Jupyter Notebook 1

  4. Reproducibility-METEOR-NLP Reproducibility-METEOR-NLP Public

    Reproducibility audit of METEOR across 73k ACL papers. Compared 8 Python/Java implementations; found up to 10-point score divergence from undocumented parameter choices. Includes Docker test harnes…

    Jupyter Notebook 2

  5. gmf-annotation-platform gmf-annotation-platform Public

    MVP pipeline for LLM-assisted annotation of AI incidents from the AIID. Imports AIID backup data, runs GPT structured-output classification against GMF taxonomy categories, and compares predictions…

    Python

  6. NLP_DeepLearning_Spring2023 NLP_DeepLearning_Spring2023 Public

    Multitask BERT fine-tuning for sentiment analysis, paraphrase detection, and semantic similarity (SST, QQP, STS-B). Explores SMART regularization, Sophia optimizer, and gradient surgery. ~5% robust…

    Python