Downloads · 30 days
0
laya-multimodal-best-all-tasks
TalalML123/laya-multimodal-best-all-tasks
laya-multimodal-best-all-tasks is a machine learning model from TalalML123. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for pytorch. The card lists the license as apache-2.0.
Joint typed-decision model (noul + score + choice) trained with RLCD.
Access
Public
Updated Sep 23, 2026
Repo size
1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt1 GB · 100%
At a glance
- Library
- pytorch
- License
- apache-2.0
- Access
- Public
- Created
- Sep 23, 2026
- Updated
- Sep 23, 2026
- SHA
- 8e36b703
Model card
From the Hugging Face model README
laya-multimodal — best all-tasks checkpoint
Joint typed-decision model (noul + score + choice) trained with RLCD.
| Hardware | AWS g5.2xlarge (NVIDIA A10G 24GB) |
| Vision | SigLIP google/siglip-base-patch16-224 (P=196, d_v=768) |
| Text | Laya / ModernBERT backbone (convaiinnovations/laya) |
| Best val accuracy | ~69.2% (stage-2, --unfreeze-text 4) |
| Code | https://github.com/TalalAhmed311/laya-multimodal |
Files
best.pt— weights (projector / cross / head / scorer + training state)history.jsonl— training logcurves.png— loss / val curves (if uploaded)
Load
from huggingface_hub import hf_hub_download
import torch
path = hf_hub_download(repo_id="TalalML123/laya-multimodal-best-all-tasks", filename="best.pt")
ckpt = torch.load(path, map_location="cpu")
# keys: projector, cross, head, scorer, ...
See the GitHub README for training / experiment details.
- Library
- pytorch
- License
- apache-2.0
- Created
- Sep 23, 2026
- Updated
- Sep 23, 2026