Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -38,7 +38,6 @@ def model_inference(
|
|
38 |
text = [text]
|
39 |
|
40 |
inputs = processor(text=text, images=images, padding=True, return_tensors="pt").to("cuda")
|
41 |
-
print("inputs",inputs)
|
42 |
|
43 |
assert decoding_strategy in [
|
44 |
"Greedy",
|
@@ -54,8 +53,8 @@ def model_inference(
|
|
54 |
generated_ids = model.generate(**inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=max_new_tokens,
|
55 |
temperature=temperature, do_sample=do_sample, repetition_penalty=repetition_penalty,
|
56 |
top_p=top_p),
|
57 |
-
generated_texts = processor.batch_decode(generated_ids, skip_special_tokens=True)
|
58 |
-
|
59 |
print("INPUT:", text, "|OUTPUT:", generated_texts)
|
60 |
return generated_texts[0]
|
61 |
|
|
|
38 |
text = [text]
|
39 |
|
40 |
inputs = processor(text=text, images=images, padding=True, return_tensors="pt").to("cuda")
|
|
|
41 |
|
42 |
assert decoding_strategy in [
|
43 |
"Greedy",
|
|
|
53 |
generated_ids = model.generate(**inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=max_new_tokens,
|
54 |
temperature=temperature, do_sample=do_sample, repetition_penalty=repetition_penalty,
|
55 |
top_p=top_p),
|
56 |
+
#generated_texts = processor.batch_decode(generated_ids, skip_special_tokens=True)
|
57 |
+
generated_texts = processor.batch_decode(generated_ids[:, inputs["input_ids"].size(1):], skip_special_tokens=True)
|
58 |
print("INPUT:", text, "|OUTPUT:", generated_texts)
|
59 |
return generated_texts[0]
|
60 |
|