Jim Lai

grimjim

sanjaisibi's profile picture

clem's profile picture

toastertech's profile picture

AI & ML interests

Experimenting primarily with 7B-12B parameter text completion models. Not all models are intended for direct end use, but aim for research and/or educational purposes. Recent Contributions: stabilized refusal direction ablation via Gram-Schmidt orthonormalization and norm-preserving interventions; confirmed reasoning transfer via model merger.

Recent Activity

new activity 8 days ago

jukofyork/creative-writing-control-vectors-v3.0:“The doom lies in yourself, not in your name.”

updated a Space 25 days ago

build-small-hackathon/dreadzone

published a Space 25 days ago

grimjim/gradiochatbotexample

View all activity

Organizations

grimjim 's collections 5

Highlighted work

My "greatest hits", sort of

grimjim/gemma-3-12b-it-norm-preserved-biprojected-abliterated

Image-Text-to-Text • 12B • Updated Jan 4 • 40 • • 26
grimjim/gemma-3-12b-it-projection-abliterated

Image-Text-to-Text • 12B • Updated Oct 30, 2025 • 14 • • 9
grimjim/gemma-3-12b-it-biprojected-abliterated

Image-Text-to-Text • 12B • Updated Nov 1, 2025 • 9 • • 7
grimjim/DeepSauerHuatuoSkywork-R1-o1-Llama-3.1-8B

Text Generation • 8B • Updated Feb 7, 2025 • 9 • 7

Quantized models

Select models helpfully quantized by others as well as myself

grimjim/meta-llama-Llama-3.2-1B-Instruct-exl2

Text Generation • Updated Nov 29, 2024 • 3
grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B-GGUF

Text Generation • 7B • Updated Aug 26, 2024 • 37 • 2
Lewdiculous/kuno-kunoichi-v1-DPO-v2-SLERP-7B-GGUF-IQ-Imatrix

7B • Updated Mar 14, 2024 • 213 • 10
grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B-8.0bpw_h8_exl2

Text Generation • Updated Aug 26, 2024 • 2

Experimental and negative results

Models that didn't always quite work out, but may still be of interest.

grimjim/franken-kunoichi-IDUS-11B

Text Generation • 11B • Updated Apr 15, 2024 • 6 • 1
grimjim/kunoichi-squared-model_stock-7B

Text Generation • 7B • Updated Apr 15, 2024 • 1
grimjim/wizard-elem-to-32k-7B

Text Generation • 7B • Updated May 27, 2024 • 10
grimjim/Llama-3-Perky-Pat-Instruct-8B

Text Generation • 8B • Updated Jul 4, 2024 • 2 • 4

Full weight models

Select models uploaded in safetensors format. Currently all are merges. Annotations here.

grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B

Text Generation • 7B • Updated May 10, 2024 • 81 • 4
grimjim/kukulemon-7B

Text Generation • 7B • Updated Mar 21, 2024 • 79 • 11
grimjim/kukulemon-spiked-9B

Text Generation • 9B • Updated Apr 27, 2024 • 6 • 3
grimjim/kukulemon-32K-7B

Text Generation • 7B • Updated Apr 27, 2024 • 11 • 1

Mirrored mergekit-ready models

Mirrored models tweaked to be more friendly for mergekit. No pickles allowed.

grimjim/llama-3-nvidia-ChatQA-1.5-8B

Text Generation • 8B • Updated May 2, 2024 • 5 • 3
grimjim/mistralai-Mistral-Nemo-Instruct-2407

Text Generation • 12B • Updated Nov 17, 2025 • 18
grimjim/mistralai-Mistral-Nemo-Base-2407

Text Generation • 12B • Updated Nov 16, 2025 • 8
grimjim/llama-3-aaditya-OpenBioLLM-8B

Text Generation • 8B • Updated May 29, 2024 • 18 •