Fill-Mask
Transformers
PyTorch
English
babylm
babylm-2026
gpt-bert
sample-efficient-pretraining
muon
custom_code
Instructions to use svsatheesh/BabySteps_SidBert-100M-mixed with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use svsatheesh/BabySteps_SidBert-100M-mixed with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="svsatheesh/BabySteps_SidBert-100M-mixed", trust_remote_code=True)# Load model directly from transformers import GPTBERTFoCausalLM model = GPTBERTFoCausalLM.from_pretrained("svsatheesh/BabySteps_SidBert-100M-mixed", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "babylm_checkpoints": true, | |
| "babylm_ckpt_dir": null, | |
| "beta1": 0.9, | |
| "beta2": 0.98, | |
| "causal_per_masked": 1, | |
| "dev_dir": "/home/sid/sidBabyLM/nanochat/data/dev", | |
| "eps": 1e-08, | |
| "full_dev": false, | |
| "grad_clip": 2.0, | |
| "hub_repo": "svsatheesh/BabySteps_ChadBert-100M-mixed", | |
| "keep_ratio": 0.1, | |
| "lr": 0.007, | |
| "mask_end": 0.15, | |
| "mask_start": 0.3, | |
| "masked_per_causal": 0, | |
| "micro_batch": 8, | |
| "muon_aux_lr": 0.007, | |
| "muon_aux_weight_decay": null, | |
| "muon_batch_updates": true, | |
| "muon_beta2": 0.95, | |
| "muon_eps2": 1e-08, | |
| "muon_lr": 0.005, | |
| "muon_momentum": 0.95, | |
| "muon_nesterov": true, | |
| "muon_ns_steps": 5, | |
| "muon_split_qkv": true, | |
| "muon_weight_decay": 0.1, | |
| "optimizer": "muon", | |
| "output_dir": "/home/sid/sidBabyLM/gptbert/runs/sweep100m/muon-100m-legacyCK-mixed-lr0.005-aux0.007-s42", | |
| "phase1_frac": 0.6, | |
| "phase2_frac": 0.8, | |
| "precision": "bf16", | |
| "push_to_hub": false, | |
| "random_ratio": 0.1, | |
| "repo": "BabyLM-community/babylm-baseline-100m-gpt-bert-mixed", | |
| "retokenize": false, | |
| "save_every": 0, | |
| "seed": 42, | |
| "tail_avg_every": 10, | |
| "tail_avg_frac": 0.0, | |
| "tokens_per_step": 131072, | |
| "total_steps": 12330, | |
| "track": "strict", | |
| "train_dir": "/home/sid/sidBabyLM/nanochat/data/strict", | |
| "val_every": 1000, | |
| "val_seq_len": 128, | |
| "val_words": 1000000, | |
| "warmup_frac": 0.016, | |
| "weight_decay": 0.1, | |
| "z_loss": 0.0001 | |
| } |