1 3

Rev

Bredvige

AI & ML interests

None yet

Recent Activity

liked a Space 23 days ago

Neogrconpet/StableGR

updated a Space about 1 month ago

Bredvige/RVC

reacted to grimjim's post with 🔥 about 1 month ago

I'm (finally) releasing a Python script that trims excess weights in Gemma2 full-weight models that bloated by ~1B parameters due to an early mergekit bug. https://github.com/jim-plus/Gemma2-mergekit-remediation I'd noticed something was off when merges of Gemma2 9B models ended up having ~10B parameters. The current mergekit package is fine, but there are still bloated models on HF that could stand to be fixed. The script assumes that it will be run from the same directory as the model weights, and will trim the unnecessary lm_head.weight tensor and corresponding index entry.

View all activity

Organizations

None yet

Bredvige's activity

liked a Space 23 days ago

StableGR

🤗

Search, download, and generate images using Civitai models

updated a Space about 1 month ago

RVC

💻

reacted to grimjim's post with 🔥 about 1 month ago

Post

2777

I'm (finally) releasing a Python script that trims excess weights in Gemma2 full-weight models that bloated by ~1B parameters due to an early mergekit bug.
https://github.com/jim-plus/Gemma2-mergekit-remediation

I'd noticed something was off when merges of Gemma2 9B models ended up having ~10B parameters. The current mergekit package is fine, but there are still bloated models on HF that could stand to be fixed.

The script assumes that it will be run from the same directory as the model weights, and will trim the unnecessary lm_head.weight tensor and corresponding index entry.