初始化项目,由ModelHub XC社区提供模型
Model: laion/swesmith-unified-1000__Qwen3-8B Source: Original Platform
This commit is contained in:
16
train_results.json
Normal file
16
train_results.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"achieved_tflops_per_gpu": 5.1867306879190735,
|
||||
"achieved_tflops_per_gpu_theoretical": 105.0250414475182,
|
||||
"epoch": 7.0,
|
||||
"loss_nan_ranks": 0,
|
||||
"loss_rank_avg": 0.04163763299584389,
|
||||
"mfu_percent": 0.3665534055066483,
|
||||
"mfu_percent_theoretical": 7.42226441325217,
|
||||
"total_flos": 1.4912561102402355e+17,
|
||||
"train_loss": 0.17323701509407588,
|
||||
"train_runtime": 898.4803,
|
||||
"train_samples_per_second": 2.781,
|
||||
"train_steps_per_second": 0.031,
|
||||
"valid_targets_mean": 7246.0,
|
||||
"valid_targets_min": 1162
|
||||
}
|
||||
Reference in New Issue
Block a user