Update README.md
Browse files
README.md
CHANGED
|
@@ -52,9 +52,6 @@ image = Image.open(requests.get(url, stream=True).raw)
|
|
| 52 |
|
| 53 |
model = PaliGemmaForConditionalGeneration.from_pretrained(
|
| 54 |
model_id,
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
torch_dtype=torch.bfloat16,
|
| 59 |
device_map=device,
|
| 60 |
).eval()
|
|
@@ -74,6 +71,7 @@ with torch.inference_mode():
|
|
| 74 |
```
|
| 75 |
|
| 76 |
|
|
|
|
| 77 |
### Results
|
| 78 |
|
| 79 |
By following the **LLM-as-Juries** evaluation method, the following results were obtained using three judge models (GPT-4o, Gemini1.5 Pro, and Claude 3.5-Sonnet). These models were evaluated based on a well-defined scoring rubric specifically designed for the VQA context, with clear criteria for each score to ensure the highest possible precision in meeting expectations.
|
|
|
|
| 52 |
|
| 53 |
model = PaliGemmaForConditionalGeneration.from_pretrained(
|
| 54 |
model_id,
|
|
|
|
|
|
|
|
|
|
| 55 |
torch_dtype=torch.bfloat16,
|
| 56 |
device_map=device,
|
| 57 |
).eval()
|
|
|
|
| 71 |
```
|
| 72 |
|
| 73 |
|
| 74 |
+
|
| 75 |
### Results
|
| 76 |
|
| 77 |
By following the **LLM-as-Juries** evaluation method, the following results were obtained using three judge models (GPT-4o, Gemini1.5 Pro, and Claude 3.5-Sonnet). These models were evaluated based on a well-defined scoring rubric specifically designed for the VQA context, with clear criteria for each score to ensure the highest possible precision in meeting expectations.
|