25 lines
778 B
JSON
25 lines
778 B
JSON
{
|
|
"model_id": "tirumalaseti/slm-125m-base",
|
|
"project": "slm-125m",
|
|
"parameter_count": 125847552,
|
|
"parameter_count_millions": 125.848,
|
|
"context_length": 1024,
|
|
"vocab_size": 16384,
|
|
"train_tokens": 2039072768,
|
|
"val_tokens": 20601856,
|
|
"total_tokens": 2059674624,
|
|
"tokens_seen": 2500329472,
|
|
"optimizer_steps": 4769,
|
|
"epochs_completed": 1.226,
|
|
"world_size": 8,
|
|
"grad_accum_steps": 2,
|
|
"final_validation_loss": 2.303484795203459,
|
|
"final_validation_perplexity": 10.00900107062853,
|
|
"reported_spend_usd": 0.0,
|
|
"reported_spend_note": "Modal billing report may lag; this value reflects the latest locally visible report.",
|
|
"corpus_mix_tokens": {
|
|
"case-law": 722081792,
|
|
"sec": 868714496,
|
|
"fineweb-edu": 468878336
|
|
}
|
|
} |