Instructions to use guygrigsby/diff-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use guygrigsby/diff-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir diff-mlx guygrigsby/diff-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
File size: 692 Bytes
77f7830 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 | {
"model": {
"dim": 768,
"n_layers": 12,
"n_heads_vanilla": 12,
"qk_head_dim": 64,
"vocab_size": 100277,
"mlp_intermediate": 2048,
"block_size": 2048,
"rope_base": 10000.0,
"rms_eps": 1e-05,
"tie_embeddings": true,
"amp_dtype": "bfloat16"
},
"train": {
"peak_lr": 0.0004,
"warmup_steps": 1000,
"total_tokens": 2000000000,
"micro_batch": 8,
"grad_accum": 4,
"weight_decay": 0.1,
"adam_beta1": 0.9,
"adam_beta2": 0.95,
"adam_eps": 1e-08,
"grad_clip": 1.0,
"eval_every": 1000,
"full_eval_every": 5000,
"monitoring_tokens": 2000000,
"full_eval_tokens": 75000000,
"save_every": 500
}
} |