Marco
commited on
Commit
·
3af94b3
1
Parent(s):
e721eb8
Corrected the readme to be compliant with the standard HF MM pipeline
Browse files
README.md
CHANGED
@@ -159,20 +159,14 @@ pixel_values = [pixel_values]
|
|
159 |
|
160 |
# generate output
|
161 |
with torch.inference_mode():
|
162 |
-
|
163 |
-
|
164 |
-
|
165 |
-
|
166 |
-
|
167 |
-
temperature=None,
|
168 |
-
repetition_penalty=None,
|
169 |
-
eos_token_id=model.generation_config.eos_token_id,
|
170 |
-
pad_token_id=text_tokenizer.pad_token_id,
|
171 |
-
use_cache=True
|
172 |
-
)
|
173 |
-
output_ids = model.generate(input_ids, pixel_values=pixel_values, attention_mask=attention_mask, **gen_kwargs)[0]
|
174 |
output = text_tokenizer.decode(output_ids, skip_special_tokens=True)
|
175 |
print(f'Output:\n{output}')
|
|
|
176 |
```
|
177 |
|
178 |
<details>
|
|
|
159 |
|
160 |
# generate output
|
161 |
with torch.inference_mode():
|
162 |
+
if inputs['pixel_values'] is not None:
|
163 |
+
inputs['pixel_values'] = [pix.to(model.dtype).to(model.device) for pix in inputs['pixel_values']]
|
164 |
+
inputs = inputs.to('cuda')
|
165 |
+
|
166 |
+
output_ids = model.generate(inputs =inputs.pop('input_ids'), **inputs)[0]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
167 |
output = text_tokenizer.decode(output_ids, skip_special_tokens=True)
|
168 |
print(f'Output:\n{output}')
|
169 |
+
|
170 |
```
|
171 |
|
172 |
<details>
|