Feature Extraction
sentence-transformers
Safetensors
Transformers
qwen2_5_omni_thinker
image-text-to-text
multimodal-embedding
Instructions to use LCO-Embedding/LCO-Embedding-Omni-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use LCO-Embedding/LCO-Embedding-Omni-7B with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("LCO-Embedding/LCO-Embedding-Omni-7B") sentences = [ "The weather is lovely today.", "It's so sunny outside!", "He drove to the stadium." ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] - Transformers
How to use LCO-Embedding/LCO-Embedding-Omni-7B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="LCO-Embedding/LCO-Embedding-Omni-7B")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForMultimodalLM tokenizer = AutoTokenizer.from_pretrained("LCO-Embedding/LCO-Embedding-Omni-7B") model = AutoModelForMultimodalLM.from_pretrained("LCO-Embedding/LCO-Embedding-Omni-7B", device_map="auto") - Notebooks
- Google Colab
- Kaggle
fix README example code bugs
Browse files
README.md
CHANGED
|
@@ -248,13 +248,13 @@ with torch.no_grad():
|
|
| 248 |
}
|
| 249 |
] for audio in batch_audios]
|
| 250 |
|
| 251 |
-
text =
|
| 252 |
messages, tokenize=False, add_generation_prompt=True
|
| 253 |
)
|
| 254 |
audio_inputs, image_inputs, video_inputs = process_mm_info(
|
| 255 |
messages, use_audio_in_video=False
|
| 256 |
)
|
| 257 |
-
inputs =
|
| 258 |
text=text,
|
| 259 |
audio=audio_inputs,
|
| 260 |
images=image_inputs,
|
|
@@ -263,7 +263,7 @@ with torch.no_grad():
|
|
| 263 |
padding=True
|
| 264 |
)
|
| 265 |
inputs = inputs.to("cuda")
|
| 266 |
-
audio_outputs =
|
| 267 |
**inputs, output_hidden_states=True, return_dict=True
|
| 268 |
).hidden_states[-1][:, -1, :]
|
| 269 |
all_audio_embeddings.append(audio_outputs.to(torch.float16).cpu())
|
|
@@ -324,13 +324,13 @@ with torch.no_grad():
|
|
| 324 |
}
|
| 325 |
] for video in batch_videos]
|
| 326 |
|
| 327 |
-
text =
|
| 328 |
messages, tokenize=False, add_generation_prompt=True
|
| 329 |
)
|
| 330 |
audio_inputs, image_inputs, video_inputs = process_mm_info(
|
| 331 |
messages, use_audio_in_video=False
|
| 332 |
)
|
| 333 |
-
inputs =
|
| 334 |
text=text,
|
| 335 |
audio=audio_inputs,
|
| 336 |
images=image_inputs,
|
|
@@ -339,7 +339,7 @@ with torch.no_grad():
|
|
| 339 |
padding=True
|
| 340 |
)
|
| 341 |
inputs = inputs.to("cuda")
|
| 342 |
-
video_outputs =
|
| 343 |
**inputs, output_hidden_states=True, return_dict=True
|
| 344 |
).hidden_states[-1][:, -1, :]
|
| 345 |
all_video_embeddings.append(video_outputs.to(torch.float16).cpu())
|
|
|
|
| 248 |
}
|
| 249 |
] for audio in batch_audios]
|
| 250 |
|
| 251 |
+
text = processor.apply_chat_template(
|
| 252 |
messages, tokenize=False, add_generation_prompt=True
|
| 253 |
)
|
| 254 |
audio_inputs, image_inputs, video_inputs = process_mm_info(
|
| 255 |
messages, use_audio_in_video=False
|
| 256 |
)
|
| 257 |
+
inputs = processor(
|
| 258 |
text=text,
|
| 259 |
audio=audio_inputs,
|
| 260 |
images=image_inputs,
|
|
|
|
| 263 |
padding=True
|
| 264 |
)
|
| 265 |
inputs = inputs.to("cuda")
|
| 266 |
+
audio_outputs = model(
|
| 267 |
**inputs, output_hidden_states=True, return_dict=True
|
| 268 |
).hidden_states[-1][:, -1, :]
|
| 269 |
all_audio_embeddings.append(audio_outputs.to(torch.float16).cpu())
|
|
|
|
| 324 |
}
|
| 325 |
] for video in batch_videos]
|
| 326 |
|
| 327 |
+
text = processor.apply_chat_template(
|
| 328 |
messages, tokenize=False, add_generation_prompt=True
|
| 329 |
)
|
| 330 |
audio_inputs, image_inputs, video_inputs = process_mm_info(
|
| 331 |
messages, use_audio_in_video=False
|
| 332 |
)
|
| 333 |
+
inputs = processor(
|
| 334 |
text=text,
|
| 335 |
audio=audio_inputs,
|
| 336 |
images=image_inputs,
|
|
|
|
| 339 |
padding=True
|
| 340 |
)
|
| 341 |
inputs = inputs.to("cuda")
|
| 342 |
+
video_outputs = model(
|
| 343 |
**inputs, output_hidden_states=True, return_dict=True
|
| 344 |
).hidden_states[-1][:, -1, :]
|
| 345 |
all_video_embeddings.append(video_outputs.to(torch.float16).cpu())
|