Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: openbmb/MiniCPM-V-4_5
|
| 4 |
+
tags:
|
| 5 |
+
- vision
|
| 6 |
+
- schematic
|
| 7 |
+
- merged
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# MiniCPM-V-4.5 (merged, schematic fine-tuned)
|
| 11 |
+
|
| 12 |
+
Full merged model (base + LoRA) for **openbmb/MiniCPM-V-4_5**, fine-tuned on PCB/schematic image descriptions. BF16.
|
| 13 |
+
|
| 14 |
+
## Load
|
| 15 |
+
|
| 16 |
+
```python
|
| 17 |
+
from transformers import AutoModel, AutoTokenizer, AutoProcessor
|
| 18 |
+
|
| 19 |
+
model = AutoModel.from_pretrained("foundation-models/minicpm-v-4.5-schematic-merged", trust_remote_code=True, torch_dtype="bfloat16")
|
| 20 |
+
tokenizer = AutoTokenizer.from_pretrained("foundation-models/minicpm-v-4.5-schematic-merged", trust_remote_code=True)
|
| 21 |
+
processor = AutoProcessor.from_pretrained("foundation-models/minicpm-v-4.5-schematic-merged", trust_remote_code=True)
|
| 22 |
+
```
|
| 23 |
+
|
| 24 |
+
## LoRA adapter
|
| 25 |
+
|
| 26 |
+
To use the adapter only: [foundation-models/minicpm-v-4.5-schematic-lora](https://huggingface.co/foundation-models/minicpm-v-4.5-schematic-lora).
|