Instructions to use geraldaton20/hd-laya with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Laya
How to use geraldaton20/hd-laya with Laya:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Commit History
Today
-
Add Download section to model card
Verified signature87cf24f -
Updated comparison chart: BANKING77, CLINC150, SNIPS, ATIS
Verified signature1d933b8 -
SNIPS+ATIS: no improvement (below/at ceiling), updated architecture
Verified signature6ef3175 -
Add SNIPS + ATIS benchmarks to HF
Verified signaturec52ddd3 -
CLINC150: all trained encoders hurt accuracy, SimHash is best
Verified signatureb3adeed -
Add CLINC150 experiment: trained HDC improves recall but not accuracy, rejected
Verified signature923ec61 -
Final architecture: base laya + SimHash HDC, reject trained approaches
Verified signaturee185536 -
Final: architecture is base laya + SimHash HDC, document trained head/encoder rejections
Verified signature5a1a76d -
Update model card with CLINC150 benchmark + index recall analysis
Verified signature01ecff9 -
Update comparison chart with CLINC150 results
Verified signatured4bed0a -
Final model card: real-world benchmarks + comparison chart
Verified signaturef354e3f -
Add comparison chart: hd-laya vs base laya
Verified signatureccbd494 -
Calibrated checkpoint (temp=1.427 for choice bucket, shape [3])
Verified signature199cb0b -
Add calibrated checkpoint (temp=1.427 for choice bucket)
Verified signature633ea71
Yesterday
-
v4 trained checkpoint: 12 epochs, +3.1pt delta on K=20
Verified signature1bf15a0 -
Withdraw artifact: prediction-identical to base
Verified signature0dd1103 -
Withdraw artifact: prediction-identical to base
Verified signature94d40eb -
Withdraw artifact: prediction-identical to base
Verified signature0b77770 -
Withdraw weights: fine-tune did not beat base; card documents the negative
Verified signaturec40e3e3 -
Model card: withdraw training-gain claim (97.9% agreement with base)
Verified signature339d21b -
Model card: real training stats, usage, and limits
Verified signature0826b09 -
Delete hd_laya_head.pt with huggingface_hub
Verified signature691144b -
Upload head.onnx.data with huggingface_hub
Verified signature8df37c8 -
Upload head.onnx with huggingface_hub
Verified signature3b68447 -
Upload hd_laya_trained.pt with huggingface_hub
Verified signaturec23e90c
2 days ago
-
Upload folder using huggingface_hub
Verified signature4d07234 -
initial commit
Verified signature7801bce