Instructions to use logitshq/Logits-Sigma-1-Duo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use logitshq/Logits-Sigma-1-Duo with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="logitshq/Logits-Sigma-1-Duo")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("logitshq/Logits-Sigma-1-Duo", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download code/serve.sh from logitshq/Logits-Sigma-1-Duo: direct link, hf CLI and curl.
- Browser
- Download file 970 Bytes
-
https://huggingface.co/logitshq/Logits-Sigma-1-Duo/resolve/main/code/serve.sh
- Command line
-
hf download hf://logitshq/Logits-Sigma-1-Duo/code/serve.sh
-
curl -L -o serve.sh https://huggingface.co/logitshq/Logits-Sigma-1-Duo/resolve/main/code/serve.sh
970 Bytes
| # Logits Sigma-1 Duo: `serve.sh <kit subcommand> [args]`, e.g. `pipeline --compact --out runs/x` or | |
| # `run --rows rows.jsonl --out out`. The engine and its one option are fixed here; Duo's settings (docs/37 §7g) live in | |
| # logits_sigma_1_duo.engine_pair.DUO. Logs the torch build, the GPU and whether causal-conv1d imports; the engine then | |
| # logs its FP8 matmul and d3-ultra's kernels, which environment.json keeps under model_source. | |
| set -euo pipefail | |
| CMD=${1:?usage: serve.sh pipeline|run [kit args]}; shift | |
| python -c "import torch; print('torch', torch.__version__, torch.cuda.get_device_name(0), torch.cuda.get_device_capability(0))" | |
| python -c "import causal_conv1d; print('causal-conv1d', causal_conv1d.__version__)" 2>/dev/null \ | |
| || echo "causal-conv1d not installed: d3-ultra uses the PyTorch reference convolution" | |
| exec python -m decision_index "$CMD" --engine logits_sigma_1_duo.engine_pair:Duo --option model="${MODEL:-/models/duo}" "$@" | |