Instructions to use Angshul/SpliNet with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Angshul/SpliNet with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="Angshul/SpliNet", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForMaskedLM model = AutoModelForMaskedLM.from_pretrained("Angshul/SpliNet", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download training_metadata.json from Angshul/SpliNet: direct link, hf CLI and curl.
- Browser
- Download file 866 Bytes
-
https://huggingface.co/Angshul/SpliNet/resolve/main/training_metadata.json
- Command line
-
hf download hf://Angshul/SpliNet/training_metadata.json
-
curl -L -o training_metadata.json https://huggingface.co/Angshul/SpliNet/resolve/main/training_metadata.json
866 Bytes
| { | |
| "model": "SpliNet", | |
| "dataset": "allenai/c4 en", | |
| "training_tokens": 2000000000, | |
| "validation_tokens": 5120000, | |
| "tokenizer_sha256": "119ec6b2af9cbbc56f297bd606b69f79f6e3130a34ee56122ee813a58d15bb9d", | |
| "vocab_size": 32000, | |
| "sequence_length": 512, | |
| "layers": 12, | |
| "hidden_size": 768, | |
| "heads": 12, | |
| "ffn_size": 3072, | |
| "spline_order": 2, | |
| "spline_radius": 16, | |
| "spline_sidedness": "single", | |
| "physical_batch": 160, | |
| "effective_batch": 1024, | |
| "optimizer": "AdamW", | |
| "base_learning_rate": 0.0001, | |
| "weight_decay": 0.01, | |
| "precision": "bfloat16", | |
| "gpu": "NVIDIA A100-SXM4-80GB", | |
| "final_validation_loss": 4.6145022583007815, | |
| "final_validation_perplexity": 100.93757518673716, | |
| "best_validation_loss": 4.6145022583007815, | |
| "openreview": "https://openreview.net/forum?id=nWHnuiEF3C", | |
| "github": "https://github.com/AngshulMajumdar/SpliNet" | |
| } |