Downloads · 30 days
0
smolvla-finetuned-20000-steps
Jonnyyy/smolvla-finetuned-20000-steps
smolvla-finetuned-20000-steps is a machine learning model from Jonnyyy. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
- Base Model: lerobot/smolvlabase - Dataset: lerobot/fmb - Training Steps: 20000 - Final Loss: 0.0200 - Training Time: 84.4 seconds
Access
Public
Updated Jul 19, 2025
Repo size
25.6 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin27.8 MB · 97%
At a glance
- Access
- Public
- Created
- Jul 19, 2025
- Updated
- Jul 19, 2025
- SHA
- 6aff9348
Model card
From the Hugging Face model README
SmolVLA Fine-tuned Model
Model Details
- Base Model: lerobot/smolvla_base
- Dataset: lerobot/fmb
- Training Steps: 20000
- Final Loss: 0.0200
- Training Time: 84.4 seconds
Training Configuration
- Batch Size: 32
- Learning Rate: 0.0001
- Optimizer: AdamW
- Scheduler: Linear with warmup
Usage
import torch
from your_model_class import SmolVLAModel
model = SmolVLAModel()
model.load_state_dict(torch.load("pytorch_model.bin"))
model.eval()
Performance
- Loss reduction: 91.2% (0.2261 → 0.0200)
- Trained on 338188 robotic manipulation samples
Model trained using Modal + Claude Code pipeline
- Created
- Jul 19, 2025
- Updated
- Jul 19, 2025