No model card yet.
License
apache-2.0
Tasks
image-text-to-text
Frameworks
transformers
Hardware-aware snippets
Runtime-specific quick starts for QuantTrio/Step3-VL-10B-AWQ. Detected hardware: No explicit hardware metadata.
Task: image-text-to-textTransformers
NVIDIA CUDA path
Best default for NVIDIA GPUs and hosted accelerator nodes.
pip install torch transformers accelerate
python - <<'PY'
from transformers import pipeline
model_id = "QuantTrio/Step3-VL-10B-AWQ"
pipe = pipeline(
task="image-text-to-text",
model=model_id,
device_map="auto",
model_kwargs={"torch_dtype": "auto"},
)
print(pipe("Hello from Inferix"))
PYModel lineage
1 reposBase model
stepfun-ai/Step3-VL-10BFinetuned
1Step3-VL-10B-AWQselected
Quantizations
0No quantization variants detected.
Model info
Namespace
QuantTrio
Visibility
public
Downloads
0
Likes
0
License: apache-2.0
Evaluation results
Entries
0
Tasks
0
Datasets
0
No evaluation results published yet.
Metadata links
Inference providers
0/0 fitNot available for inference yet.
Spaces using this model
No spaces linked yet.
Part of collections
No public collections include this model.
Tags
importedhuggingfaceminiotransformerssafetensorsstep_roboticstext-generationvLLMAWQimage-text-to-textconversationalcustom_codearxiv:2601.09668base_model:stepfun-ai/Step3-VL-10Bbase_model:quantized:stepfun-ai/Step3-VL-10Blicense:apache-2.0endpoints_compatible4-bitawqregion:us