Skip to content

Commit

Permalink
Update data and shampoo sub name
Browse files Browse the repository at this point in the history
  • Loading branch information
@fsschneider
fsschneider committed Sep 30, 2024
1 parent 202104b commit 6ce8ff9
Show file tree
Hide file tree
Showing 8 changed files with 851 additions and 1,550 deletions.
1,115 changes: 503 additions & 612 deletions scoring/plotting_correlations.ipynb
View file

Large diffs are not rendered by default.

32 changes: 8 additions & 24 deletions scoring/plotting_identify_bugged_runs.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,7 @@
},
{
"cell_type": "code",
"execution_count": 2,
"execution_count": 12,
"metadata": {},
"outputs": [
{
Expand Down Expand Up @@ -97,22 +97,6 @@
"['wmt_glu_tanh_jax', 'fastmri_jax', 'criteo1tb_jax', 'librispeech_conformer_layernorm_jax', 'fastmri_tanh_jax', 'ogbg_jax', 'ogbg_model_size_jax', 'imagenet_vit_jax', 'criteo1tb_embed_init_jax', 'librispeech_conformer_jax', 'librispeech_deepspeech_jax', 'wmt_jax', 'imagenet_resnet_large_bn_init_jax', 'imagenet_resnet_jax']\n",
"['wmt_glu_tanh_jax', 'fastmri_jax', 'criteo1tb_jax', 'librispeech_conformer_layernorm_jax', 'fastmri_tanh_jax', 'ogbg_jax', 'ogbg_model_size_jax', 'imagenet_vit_jax', 'criteo1tb_embed_init_jax', 'librispeech_conformer_jax', 'librispeech_deepspeech_jax', 'wmt_jax', 'imagenet_resnet_large_bn_init_jax', 'imagenet_resnet_jax']\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"2024-09-25 14:51:50.916219: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n",
"/home/frank/anaconda3/envs/algoperf_pt/lib/python3.9/site-packages/tensorflow_addons/utils/tfa_eol_msg.py:23: UserWarning: \n",
"\n",
"TensorFlow Addons (TFA) has ended development and introduction of new features.\n",
"TFA has entered a minimal maintenance and release mode until a planned end of life in May 2024.\n",
"Please modify downstream libraries to take dependencies from other repositories in our TensorFlow community (e.g. Keras, Keras-CV, and Keras-NLP). \n",
"\n",
"For more information see: https://github.com/tensorflow/addons/issues/2807 \n",
"\n",
" warnings.warn(\n"
]
}
],
"source": [
Expand All @@ -130,7 +114,7 @@
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 13,
"metadata": {},
"outputs": [],
"source": [
Expand All @@ -142,7 +126,7 @@
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": 14,
"metadata": {},
"outputs": [
{
Expand All @@ -152,16 +136,16 @@
"Number of runs that need to be rerun for shampoo_submission: 0\n",
"Number of runs that need to be rerun for prize_qualification_baseline: 6\n",
"Number of runs that need to be rerun for caspr_adaptive: 0\n",
"Number of runs that need to be rerun for schedule_free_adamw: 2\n",
"Number of runs that need to be rerun for schedule_free_adamw: 0\n",
"Number of runs that need to be rerun for schedule_free_prodigy: 0\n",
"Number of runs that need to be rerun for amos: 9\n",
"Number of runs that need to be rerun for amos: 0\n",
"Number of runs that need to be rerun for lawa_ema: 0\n",
"Number of runs that need to be rerun for lawa_queue: 0\n",
"Number of runs that need to be rerun for cyclic_lr: 25\n",
"Number of runs that need to be rerun for cyclic_lr: 0\n",
"Number of runs that need to be rerun for generalized_adam: 0\n",
"Number of runs that need to be rerun for nadamp: 7\n",
"Percentage of runs that stopped early: 14.59% (533)\n",
"Percentage of runs that need to be rerun: 1.34% (49)\n"
"Percentage of runs that stopped early: 13.55% (495)\n",
"Percentage of runs that need to be rerun: 0.36% (13)\n"
]
}
],
Expand Down
100 changes: 18 additions & 82 deletions scoring/plotting_performance_profiles.ipynb
View file

Large diffs are not rendered by default.

936 changes: 270 additions & 666 deletions scoring/plotting_runtimes.ipynb
View file

Large diffs are not rendered by default.

88 changes: 15 additions & 73 deletions scoring/plotting_speedups.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,18 +2,9 @@
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/frank/anaconda3/envs/algoperf_pt/lib/python3.9/site-packages/scipy/__init__.py:155: UserWarning: A NumPy version >=1.18.5 and <1.25.0 is required for this version of SciPy (detected version 1.26.3\n",
" warnings.warn(f\"A NumPy version >={np_minversion} and <{np_maxversion}\"\n"
]
}
],
"outputs": [],
"source": [
"import pandas as pd\n",
"import numpy as np\n",
Expand All @@ -27,12 +18,12 @@
},
{
"cell_type": "code",
"execution_count": 2,
"execution_count": 7,
"metadata": {},
"outputs": [],
"source": [
"# Load results from disk\n",
"SELF_TUNING = False\n",
"SELF_TUNING = True\n",
"results_path = '../../results/'\n",
"file_path = 'self_tuning/self_tuning_scoring_results.txt' if SELF_TUNING else 'external_tuning/external_tuning_scoring_results.txt'\n",
"RESULTS_FILE = os.path.join(results_path, file_path)\n",
Expand All @@ -43,60 +34,16 @@
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"2024-09-25 14:43:03.499640: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n",
"/home/frank/anaconda3/envs/algoperf_pt/lib/python3.9/site-packages/tensorflow_addons/utils/tfa_eol_msg.py:23: UserWarning: \n",
"\n",
"TensorFlow Addons (TFA) has ended development and introduction of new features.\n",
"TFA has entered a minimal maintenance and release mode until a planned end of life in May 2024.\n",
"Please modify downstream libraries to take dependencies from other repositories in our TensorFlow community (e.g. Keras, Keras-CV, and Keras-NLP). \n",
"\n",
"For more information see: https://github.com/tensorflow/addons/issues/2807 \n",
"\n",
" warnings.warn(\n",
"WARNING:absl:Expecting 14 workloads but found 18 workloads for prize_qualification_baseline.\n",
"WARNING:absl:Expecting 5 studies for workload criteo1tb_layernorm_jax but found 3 studies for prize_qualification_baseline.\n",
"WARNING:absl:Expecting 5 studies for workload fastmri_model_size_jax but found 3 studies for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload fastmri_model_size_jax but found 4 trials for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload fastmri_model_size_jax but found 4 trials for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_4: Expecting 5 trials for workload fastmri_model_size_jax but found 4 trials for prize_qualification_baseline.\n",
"WARNING:absl:Expecting 5 studies for workload imagenet_resnet_silu_jax but found 4 studies for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload imagenet_resnet_silu_jax but found 1 trials for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload imagenet_resnet_silu_jax but found 1 trials for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload imagenet_resnet_silu_jax but found 1 trials for prize_qualification_baseline.\n",
"WARNING:absl:In Study study_4: Expecting 5 trials for workload imagenet_resnet_silu_jax but found 1 trials for prize_qualification_baseline.\n",
"WARNING:absl:Expecting 5 studies for workload ogbg_gelu_jax but found 3 studies for prize_qualification_baseline.\n",
"WARNING:absl:Expecting 14 workloads but found 11 workloads for caspr_adaptive.\n",
"WARNING:absl:Expecting 14 workloads but found 12 workloads for schedule_free_prodigy.\n",
"WARNING:absl:In Study study_0: Expecting 5 trials for workload criteo1tb_embed_init_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload criteo1tb_embed_init_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload criteo1tb_embed_init_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload criteo1tb_embed_init_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_4: Expecting 5 trials for workload criteo1tb_embed_init_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_0: Expecting 5 trials for workload criteo1tb_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload criteo1tb_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload criteo1tb_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload criteo1tb_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_4: Expecting 5 trials for workload criteo1tb_jax but found 1 trials for amos.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 4 trials for lawa_ema.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 3 trials for lawa_ema.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 4 trials for lawa_ema.\n",
"WARNING:absl:In Study study_4: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 4 trials for lawa_ema.\n",
"WARNING:absl:Expecting 5 studies for workload librispeech_deepspeech_pytorch but found 3 studies for lawa_ema.\n",
"WARNING:absl:In Study study_0: Expecting 5 trials for workload librispeech_deepspeech_pytorch but found 1 trials for lawa_ema.\n",
"WARNING:absl:In Study study_1: Expecting 5 trials for workload librispeech_deepspeech_pytorch but found 1 trials for lawa_ema.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload librispeech_deepspeech_pytorch but found 2 trials for lawa_ema.\n",
"WARNING:absl:In Study study_0: Expecting 5 trials for workload criteo1tb_embed_init_pytorch but found 4 trials for lawa_queue.\n",
"WARNING:absl:In Study study_0: Expecting 5 trials for workload criteo1tb_pytorch but found 4 trials for lawa_queue.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload imagenet_vit_pytorch but found 4 trials for lawa_queue.\n",
"WARNING:absl:In Study study_2: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 4 trials for lawa_queue.\n",
"WARNING:absl:In Study study_3: Expecting 5 trials for workload librispeech_conformer_layernorm_pytorch but found 4 trials for lawa_queue.\n"
"WARNING:absl:Expecting 5 studies for workload librispeech_conformer_layernorm_pytorch but found 4 studies for AdamG.\n",
"WARNING:absl:Expecting 14 workloads but found 12 workloads for sinv6_75.\n",
"WARNING:absl:Expecting 14 workloads but found 12 workloads for sinv6.\n"
]
}
],
Expand All @@ -114,7 +61,7 @@
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": 9,
"metadata": {},
"outputs": [],
"source": [
Expand Down Expand Up @@ -149,7 +96,7 @@
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 10,
"metadata": {},
"outputs": [
{
Expand All @@ -161,17 +108,12 @@
"workload & speedup \\\\\n",
"submission & \\\\\n",
"\\midrule\n",
"\\shampoosub & +27.87\\% \\\\\n",
"\\sfadam & +26.60\\% \\\\\n",
"\\caspr & +24.33\\% \\\\\n",
"\\generalizedadam & +10.89\\% \\\\\n",
"\\cycliclr & +10.67\\% \\\\\n",
"\\lawaq & +7.98\\% \\\\\n",
"\\nadamp & +3.37\\% \\\\\n",
"\\amos & +1.46\\% \\\\\n",
"\\sfadam & +7.76\\% \\\\\n",
"\\baseline & -0.00\\% \\\\\n",
"\\lawaema & -9.17\\% \\\\\n",
"\\sfprodigy & -15.69\\% \\\\\n",
"\\nadamwseq & -92.44\\% \\\\\n",
"\\sinvnum & -157.67\\% \\\\\n",
"\\sinv & -168.63\\% \\\\\n",
"\\adamg & -294.17\\% \\\\\n",
"\\bottomrule\n",
"\\end{tabular}\n",
"\n"
Expand Down
Loading

0 comments on commit 6ce8ff9

Please sign in to comment.