fastai

Capecape's group workspace

Group: timm

4685

107

1-4

of 4

Tags

Notes

Author

capecape

State

Finished

Start time

June 11th, 2022 1:37:19 AM

Sweep

f6vrn8k3

Runtime

14m 55s

Tracked hours

4m 51s

Run path

fastai/fine_tune_timm/blgqshla

Linux-5.13.0-48-generic-x86_64-with-glibc2.31

Python version

3.9.10

Git repository

git clone git@github.com:tcapelle/fastai_timm.git

Git state

git checkout -b "silver-sweep-69" 2170e4ac850f79922b54a886c5595636823795af

Command

/home/jhoward/git/ext/fastai_timm/fine_tune.py --dataset=PETS --learning_rate=0.008 --model_name=regnetz_e8 --num_experiments=3 --pool=concat --resize_method=squish

System Hardware

CPU count	64
GPU count	3
GPU type	NVIDIA GeForce RTX 3090

W&B CLI Version

0.12.17

Group

timm

Config parameters are your model's inputs. Learn more

▶
Config parameters:{} 35 keys
- batch per epoch:
  184
- batch size:
  32
- batch_size:
  32
- dataset:
  "PETS"
- dataset.tfms:
  "[Pipeline: PILBase.create, Pipeline: partial -> Categorize -- {'vocab': None, 'sort': True, 'add_na': False}]"
- device:
  "cuda"
- dls.after_batch:
  "Pipeline: IntToFloatTensor -- {'div': 255.0, 'div_mask': 1}"
- dls.after_item:
  "Pipeline: Resize -- {'size': (224, 224), 'method': 'squish', 'pad_mode': 'reflection', 'resamples': (<Resampling.BILINEAR: 2>, 0), 'p': 1.0} -> ToTensor"
- dls.before_batch:
  "Pipeline: "
- epochs:
  5
- frozen:
  false
- frozen idx:
  0
- img_size:
  224
- input 1 dim 1:
  32
- input 1 dim 2:
  3
- input 1 dim 3:
  224
- input 1 dim 4:
  224
- ▶
  Learner:{} 16 keys
- learning_rate:
  0.008
- MixedPrecision:
  true
- model parameters:
  57,774,488
- model_name:
  "regnetz_e8"
- n_inp:
  1
- num_experiments:
  3
- ParamScheduler:
  true
- pool:
  "concat"
- ProgressCallback:
  true
- ▶
  Recorder:{} 3 keys
  - add_time:
    true
  - train_metrics:
    false
  - valid_metrics:
    true
- resize_method:
  "squish"
- seed:
  42
- split_func:
  "default"
- TrainEvalCallback:
  true
- wandb_entity:
  "capecape"
- wandb_project:
  "fine_tune_timm"
- ▶
  WandbCallback:{} 11 keys

Summary metrics are your model's outputs. Learn more

▶
Summary metrics:{} 20 keys
- accuracy:
  0.925575077533722
- epoch:
  6
- eps_0:
  0.00001
- eps_1:
  0.00001
- error_rate:
  0.07442490011453629
- fit_time:
  283.6589052710042
- GPU_mem:
  8.341796875
- lr_0:
  0.00000000063795977363
- lr_1:
  0.00000006379597736281
- model_family:
  "regnetz"
- mom_0:
  0.9499994050946168
- mom_1:
  0.9499994050946168
- raw_loss:
  0.0037908940576016903
- sqr_mom_0:
  0.99
- sqr_mom_1:
  0.99
- train_loss:
  0.057738788425922394
- train_samples_per_sec:
  136.6662046636687
- valid_loss:
  0.2867913842201233
- wd_0:
  0.01
- wd_1:
  0.01

This run produced these artifacts as outputs. Total: 1. Learn more

wandb-history

run-blgqshla-history:v0