law-game-evidence-replacement-finetune

• eval_loss: 3.4253
• eval_map: 0.8264
• eval_map_50: 0.8488
• eval_map_75: 0.8441
• eval_map_small: 0.5688
• eval_map_medium: 0.9527
• eval_map_large: 0.8547
• eval_mar_1: 0.7043
• eval_mar_10: 0.9575
• eval_mar_100: 0.9727
• eval_mar_small: 0.5949
• eval_mar_medium: 0.9738
• eval_mar_large: 0.9894
• eval_map_evidence: -1.0
• eval_mar_100_evidence: -1.0
• eval_map_ambulance: 0.9802
• eval_mar_100_ambulance: 0.9899
• eval_map_artificial_target: 0.9267
• eval_mar_100_artificial_target: 0.9572
• eval_map_cartridge: 0.9742
• eval_mar_100_cartridge: 0.9949
• eval_map_gun: 0.9165
• eval_mar_100_gun: 0.9403
• eval_map_knife: 0.8599
• eval_mar_100_knife: 0.931
• eval_map_police: 0.9935
• eval_mar_100_police: 0.9959
• eval_map_traffic: 0.9586
• eval_mar_100_traffic: 0.9726
• eval_map_traffic_cone: 0.0013
• eval_mar_100_traffic_cone: 1.0
• eval_runtime: 50.5267
• eval_samples_per_second: 16.467
• eval_steps_per_second: 2.058
• epoch: 28.0
• step: 5152

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

• learning_rate: 5e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 300
• num_epochs: 50

Framework versions

• Transformers 4.44.0.dev0
• Pytorch 2.3.1+cu121
• Tokenizers 0.19.1