Create meta.yaml
Browse files
meta.yaml
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
license: Apache License 2.0
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen2.5-VL-3B-Instruct
|
| 4 |
+
- Qwen/Qwen2.5-VL-7B-Instruct
|
| 5 |
+
- lmms-lab/llava-onevision-qwen2-7b-ov
|
| 6 |
+
tags:
|
| 7 |
+
- robotics
|
| 8 |
+
- vision-language-action-model
|
| 9 |
+
- vision-language-model
|
| 10 |
+
library_name: transformers
|
| 11 |
+
|
| 12 |
+
# Collection Metadata (Referencing InternRobotics/VLN-PE style)
|
| 13 |
+
repo: InternRobotics/RoboInter-VLM
|
| 14 |
+
type: "checkpoint-collection"
|
| 15 |
+
description: "Collection of RoboInterVLM checkpoints and configs fine-tuned on RoboInter-VQA."
|
| 16 |
+
checkpoints:
|
| 17 |
+
- name: RoboInterVLM_qwenvl25_3b
|
| 18 |
+
path: RoboInterVLM_qwenvl25_3b/
|
| 19 |
+
notes: "Lightweight Qwen2.5-VL model"
|
| 20 |
+
- name: RoboInterVLM_qwenvl25_7b
|
| 21 |
+
path: RoboInterVLM_qwenvl25_7b/
|
| 22 |
+
notes: "Stronger performance Qwen2.5-VL backbone"
|
| 23 |
+
- name: RoboInterVLM_llava_one_vision_7B
|
| 24 |
+
path: RoboInterVLM_llava_one_vision_7B/
|
| 25 |
+
notes: "LLaVA-OneVision (SigLIP + Qwen2) backbone"
|