CHOI0126/llava-c3-stage2
llava-c3-stage2 — LLaVA-1.5 Mix-665K repacked for the C3 (QK-norm) experiment Stage 2 (full instruction tuning) data for the C3 architectural-causality QK-norm injection experiment in minsik-choi126/mllm-model-merging. The bundle wraps the upstream LLaVA-1.5 Mix-665K JSON together with all five image sources (COCO 2017 train, GQA, Visual Genome, OCR-VQA, TextVQA) in the directory layout training/configs/data/stage2_llava_mix665k.json expects, plus a single setup.sh to do all… See the full description on the dataset page: https://huggingface.co/datasets/CHOI0126/llava-c3-stage2.
| Author / organization | CHOI0126 |
|---|---|
| License (as declared) | other |
| Downloads | 331 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 20 May 2026 |
| Last updated at source | 20 May 2026 |
Popularity, last 3 days
331 now (+2 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Recent changes
- 16 Oct 2026 Description: llava-c3-stage2 — LLaVA-1.5 Mix-665K repacked for the C3 (QK-norm) experiment Stage 2 (full instruction tuning) data for the C3 architectural-causality QK-norm injection experiment in minsik-choi126/mllm-model-merging. The bundle wraps the → llava-c3-stage2 — LLaVA-1.5 Mix-665K repacked for the C3 (QK-norm) experiment Stage 2 (full instruction tuning) data for the C3 architectural-causality QK-norm injection experiment in minsik-choi126/mllm-model-merging. The bundle wraps the
Changes detected by Silytics between refreshes of the source.
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.