Downloads · 30 days
7
23% of all-time downloads
internvl3-8b-robovqa-stage2
agiri123/internvl3-8b-robovqa-stage2
internvl3-8b-robovqa-stage2 is a robotics model from agiri123. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for peft. The card lists the license as mit.
QLoRA adapter for InternVL3-8B fine-tuned on RoboVQA for multi-turn robotics reasoning.
All-time downloads
31
Public
Repo size
3.9 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt2.6 GB · 67%
At a glance
- Task
- Robotics
- Library
- peft
- License
- mit
- Access
- Public
- Created
- Feb 6, 2026
- Updated
- Feb 6, 2026
- SHA
- 79dc37f6
Base models
Model card
From the Hugging Face model README
InternVL3-8B-RoboVQA-Stage2
QLoRA adapter for InternVL3-8B fine-tuned on RoboVQA for multi-turn robotics reasoning.
Overview
| Base Model | OpenGVLab/InternVL3-8B |
| Training Data | 199,721 multi-turn conversations from RoboVQA |
| Task | Multi-turn reasoning (Stage 2 of 2-stage curriculum) |
| Eval Loss | 0.012 |
| Perplexity | 1.012 |
Training
This is Stage 2 of a two-stage curriculum learning approach:
- Stage 1: Single QA pairs → visual grounding
- Stage 2 (this model): Multi-turn conversations → reasoning chains
Initialized from Stage 1 checkpoint, trained on DGX Spark (GB10 Blackwell, 128GB unified memory) for ~17 days.
Usage
from peft import PeftModel
from transformers import AutoModel
base_model = AutoModel.from_pretrained("OpenGVLab/InternVL3-8B", trust_remote_code=True)
model = PeftModel.from_pretrained(base_model, "agiri123/internvl3-8b-robovqa-stage2")
Links
- Repository: github.com/giricme/vlm-ft
- Stage 1 Model: agiri123/internvl3-8b-robovqa-stage1
See the repository README for full training details, hyperparameters, and curriculum validation experiments.
- Task
- Robotics
- Library
- peft
- License
- mit
- Languages
- en
- Created
- Feb 6, 2026
- Updated
- Feb 6, 2026