Instructions to use Zero-Vision/Llama-3-MixSenseV1_1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Zero-Vision/Llama-3-MixSenseV1_1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Zero-Vision/Llama-3-MixSenseV1_1", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Zero-Vision/Llama-3-MixSenseV1_1", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Zero-Vision/Llama-3-MixSenseV1_1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Zero-Vision/Llama-3-MixSenseV1_1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Zero-Vision/Llama-3-MixSenseV1_1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Zero-Vision/Llama-3-MixSenseV1_1

SGLang

How to use Zero-Vision/Llama-3-MixSenseV1_1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Zero-Vision/Llama-3-MixSenseV1_1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Zero-Vision/Llama-3-MixSenseV1_1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Zero-Vision/Llama-3-MixSenseV1_1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Zero-Vision/Llama-3-MixSenseV1_1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Zero-Vision/Llama-3-MixSenseV1_1 with Docker Model Runner:
```
docker model run hf.co/Zero-Vision/Llama-3-MixSenseV1_1
```

Todokete commited on Apr 19, 2025

Commit

a044327

verified ·

1 Parent(s): c3314a5

Fix: Handle cache_position argument for newer Transformers

Browse files

Fixes "TypeError: MixsenseLlamaForCausalLM.forward() got an unexpected keyword argument 'cache_position'" in newer Transformers releases by adding cache_position to the forward and prepare_inputs_for_generation method signatures and passing it to the super() calls.

Files changed (1) hide show

modeling_mixsense_llama.py +4 -1

modeling_mixsense_llama.py CHANGED Viewed

@@ -1115,6 +1115,7 @@ class MixsenseLlamaForCausalLM(LlamaForCausalLM, MixsenseMetaForCausalLM):
         images: Optional[torch.FloatTensor] = None,
         image_sizes: Optional[List[List[int]]] = None,
         return_dict: Optional[bool] = None,
     ) -> Union[Tuple, CausalLMOutputWithPast]:
         if inputs_embeds is None:
             (
@@ -1144,6 +1145,7 @@ class MixsenseLlamaForCausalLM(LlamaForCausalLM, MixsenseMetaForCausalLM):
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
             return_dict=return_dict,
         )
     @torch.no_grad()
@@ -1181,7 +1183,7 @@ class MixsenseLlamaForCausalLM(LlamaForCausalLM, MixsenseMetaForCausalLM):
         return output
     def prepare_inputs_for_generation(
-        self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs
     ):
         images = kwargs.pop("images", None)
         image_sizes = kwargs.pop("image_sizes", None)
@@ -1189,6 +1191,7 @@ class MixsenseLlamaForCausalLM(LlamaForCausalLM, MixsenseMetaForCausalLM):
             input_ids,
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
             **kwargs,
         )
         if images is not None:

         images: Optional[torch.FloatTensor] = None,
         image_sizes: Optional[List[List[int]]] = None,
         return_dict: Optional[bool] = None,
+        cache_position: Optional[torch.LongTensor] = None,
     ) -> Union[Tuple, CausalLMOutputWithPast]:
         if inputs_embeds is None:
             (
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
             return_dict=return_dict,
+            cache_position=cache_position,
         )
     @torch.no_grad()
         return output
     def prepare_inputs_for_generation(
+        self, input_ids, past_key_values=None, inputs_embeds=None, cache_position=None, **kwargs
     ):
         images = kwargs.pop("images", None)
         image_sizes = kwargs.pop("image_sizes", None)
             input_ids,
             past_key_values=past_key_values,
             inputs_embeds=inputs_embeds,
+            cache_position=cache_position,
             **kwargs,
         )
         if images is not None: