qwen2.5_1.5b_500k_16kcw_2ep_armv8 / train_results.json
ahmedheakl's picture
End of training
5d3fa16 verified
raw
history blame contribute delete
227 Bytes
{
"epoch": 1.9999830999721149,
"total_flos": 1.4764570273982185e+19,
"train_loss": 0.003653812557283815,
"train_runtime": 130211.2924,
"train_samples_per_second": 7.271,
"train_steps_per_second": 0.909
}