Molbap/portal-repro-results
PORTAL reproduction — lmms-eval results Raw lmms-eval 0.7.3 outputs (results JSON + per-sample logs) for the minimal reproduction of "Rethinking Projector Training in Multimodal LLMs" (arXiv: 2610.18598). llava-qwen25-05b-portal-stage2/ — PORTAL projector (training-free, closed form) llava-qwen25-05b-random-stage2/ — Random projector control Both: CLIP ViT-L/14-336 (frozen) + 2-layer MLP-GELU projector + Qwen2.5-0.5B-Instruct, identical Stage-II (60k LLaVA-instruct subset, lr… See the full description on the dataset page: https://huggingface.co/datasets/Molbap/portal-repro-results.
| Author / organization | Molbap |
|---|---|
| License (as declared) | cc-by-4.0 |
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 16 October 2026 |
| Last updated at source | 16 October 2026 |
Connected in the knowledge graph
- Rethinking Projector Training in Multimodal LLMsPaper · cites
Recent changes
- 16 Oct 2026 Description: none → PORTAL reproduction — lmms-eval results Raw lmms-eval 0.7.3 outputs (results JSON + per-sample logs) for the minimal reproduction of "Rethinking Projector Training in Multimodal LLMs" (arXiv: 2610.18598). llava-qwen25-05b-portal-stage2/ — P
- 16 Oct 2026 Topics added ml-intern, evaluation, multimodal
- 16 Oct 2026 License: none → cc-by-4.0
Changes detected by Silytics between refreshes of the source.
Related papers
More from Molbap
- Molbap/hf_cached_embeds_logDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.