TinyCeNN-LM-Distilled / student_config.json
vtava's picture
Upload CeNN-only distilled student (10,002,432 tokens)
1c1b878 verified
Raw History Blame Contribute Delete
437 Bytes
{
"format_version": 1,
"architecture": "cenn-only-replacement",
"base_model": "arnir0/Tiny-LLM",
"layer_indices": [
0
],
"cenn": {
"hidden_size": 192,
"kernel_size": 3,
"expansion": 4,
"steps": 7,
"dilations": [
1,
2,
4,
8,
16,
32,
64
],
"rms_norm_eps": 1e-05,
"dropout": 0.0
},
"training": {
"distilled": true,
"tokens": 9420800
}
}