Skip to main content
ebs
Projects
test-2172
Workspace
Log in
Sign up
Overview
Workspace
Runs
Automat.
Sweeps
Reports
Artifacts
Ebs's workspace
Personal workspace
Automated workspace
Changes are only visible to you.
Runs
23
Name
2 visualized
clip-grad-norm-rank0
clip-grad-norm-rank0
clip-grad-norm-full-tensor
clip-grad-norm-full-tensor
desert-pyramid-29
desert-pyramid-29
true-tree-28
true-tree-28
splendid-monkey-27
splendid-monkey-27
llama32-1b-lora-distributed-unscaled-largelr-cosscheduler
llama32-1b-lora-distributed-unscaled-largelr-cosscheduler
llama32-1b-lora-distributed-largescaler-largelr-cosscheduler
llama32-1b-lora-distributed-largescaler-largelr-cosscheduler
llama32-1b-lora-distributed-largescaler-largelr
llama32-1b-lora-distributed-largescaler-largelr
llama32-1b-lora-distributed-noscaler-largelr
llama32-1b-lora-distributed-noscaler-largelr
llama32-1b-lora-single-noscaler-largelr
llama32-1b-lora-single-noscaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-lora-single-largescaler-largelr
llama32-1b-singledevice-scaled
llama32-1b-singledevice-scaled
llama32-1b-singledevice-unscaled
llama32-1b-singledevice-unscaled
llama32-1b-singledevice-scaled
llama32-1b-singledevice-scaled
llama32-1b-large-scaler-adam-unfused
llama32-1b-large-scaler-adam-unfused
llama32-1b-baseline-adam-unfused
llama32-1b-baseline-adam-unfused
llama32-1b-large-scaler-sgd
llama32-1b-large-scaler-sgd
llama32-1b-baseline-sgd
llama32-1b-baseline-sgd
1-20
of 23
Settings
Add panels
Charts
8
1-6 of 8
tokens_per_second_per_gpu
tokens_per_second_per_gpu
20
40
60
80
100
global_step
1000
1500
2000
2500
clip-grad-norm-rank0
clip-grad-norm-full-tensor
peak_memory_reserved
peak_memory_reserved
20
40
60
80
100
global_step
5
6
7
clip-grad-norm-rank0
clip-grad-norm-full-tensor
peak_memory_alloc
peak_memory_alloc
20
40
60
80
100
global_step
3
4
5
clip-grad-norm-rank0
clip-grad-norm-full-tensor
peak_memory_active
peak_memory_active
20
40
60
80
100
global_step
3
4
5
clip-grad-norm-rank0
clip-grad-norm-full-tensor
lr
lr
20
40
60
80
100
global_step
0.00005
0.0001
0.00015
0.0002
0.00025
clip-grad-norm-rank0
clip-grad-norm-full-tensor
loss
loss
20
40
60
80
100
global_step
1.2
1.4
1.6
1.8
2
clip-grad-norm-rank0
clip-grad-norm-full-tensor
System
31
1-6 of 31
Add section