28 lines
608 B
JSON
28 lines
608 B
JSON
|
{
|
||
|
|
"method": "scheduled_qat_linear",
|
||
|
|
"model": "HuggingFaceTB/SmolLM2-1.7B",
|
||
|
|
"target_bits": 4,
|
||
|
|
"num_cores": 8,
|
||
|
|
"wikitext103": {
|
||
|
|
"test_loss": 3.0391995906829834,
|
||
|
|
"test_perplexity": 20.88851717837168
|
||
|
|
},
|
||
|
|
"training": {
|
||
|
|
"total_steps": 186,
|
||
|
|
"elapsed_seconds": 1149.5,
|
||
|
|
"per_core_batch": 1,
|
||
|
|
"grad_accum_steps": 8,
|
||
|
|
"global_eff_batch": 64,
|
||
|
|
"seq_length": 512,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"epochs": 3,
|
||
|
|
"final_bits": 4
|
||
|
|
},
|
||
|
|
"schedule": {
|
||
|
|
"type": "linear",
|
||
|
|
"warmup_epochs": 0.1,
|
||
|
|
"stabilization_epochs": 0.3,
|
||
|
|
"start_bits": 32,
|
||
|
|
"target_bits": 4
|
||
|
|
}
|
||
|
|
}
|