mlcommons training_policies issues

mlcommons / training_policies

Issues related to MLPerf™ training policies, including rules and suggested changes

https://mlcommons.org/en/groups/training

Apache License 2.0

93 stars 66 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Add new Floating Point Formats for Submission

#300 bitfort closed 4 years ago
1
Update/change documentation for v0.7

#299 TheKanter closed 4 years ago
3
Clarification that "noise_shape" in Dropout is Tunable

#298 bitfort closed 4 years ago
5
Number of Runs for v0.7

#297 bitfort closed 4 years ago
4
Convergence issue with MXNet ResNet-50 using Lars with 4K Batch size

#296 rnaidu02 closed 4 years ago
4
Relax the rule in the context of COVID

#295 danielcdh closed 4 years ago
2
How many runs are required for the new benchmarks, e.g., BERT, DLRM?

#294 petermattson closed 4 years ago
1
Finalization of Logging and Compliance

#293 bitfort closed 4 years ago
2
Clarify Bert Stopping Criteria

#292 bitfort opened 4 years ago
3
DLRM - custom embedding kernels should be allowed

#291 jonathan-cohen-nvidia opened 4 years ago
7
Delete Deep Speech from the table.

#290 bitfort closed 4 years ago
0
DeepSpeech 2 isn't a benchmark

#289 TheKanter closed 4 years ago
2
DLRM - Exploring current optimizer for convergence with large batch sizes (512K)

#288 rnaidu02 closed 4 years ago
2
SSD Eval Frequency

#287 bitfort opened 4 years ago
3
Minigo real games played for each model should be logged and consider a tunnable hyperparameter

#286 rnaidu02 closed 4 years ago
3
Quantize Allowed Batch Sizes

#285 bitfort opened 4 years ago
1
Bar For HP Table Modification

#284 bitfort opened 4 years ago
2
MiniGo: Allow Tensorflow 2.0 patches in Tensorflow 1.15 within minigo

#283 rnaidu02 closed 4 years ago
2
Hyperparameter Plan for v0.7

#282 bitfort closed 4 years ago
4
Allow available software to be released before publication, not submission

#281 jonathan-cohen-nvidia opened 4 years ago
3
Rules on eval of Mini Go target CP.

#280 petermattson closed 4 years ago
1
Potential delay in the training submission deadline?

#279 christ1ne closed 4 years ago
4
Proposal from the single chip working group

#278 bitfort opened 4 years ago
1
Updated rules to specify which optimizers are allowed per benchmark

#277 jonathan-cohen-nvidia closed 4 years ago
5
Re-examine embedding table size for DLRM

#276 bitfort opened 4 years ago
0
Notification of Intention to Submit

#275 bitfort opened 4 years ago
2
adding checkpoint related rules to the training rules

#274 christ1ne opened 4 years ago
1
Investigate allowing async training for DLRM

#273 bitfort opened 4 years ago
5
Resnet for larger scales

#272 bitfort closed 4 years ago
4
Moving away from white-list/black-list?

#271 bitfort opened 4 years ago
0
Resnet Learning Rate Schedule

#270 bitfort closed 4 years ago
3
Submission Timeline

#269 bitfort closed 4 years ago
10
Transparency, review and governance questions

#268 bitfort opened 4 years ago
0
Audit Process Rules

#267 bitfort closed 4 years ago
2
Adam Equivalence

#266 bitfort closed 4 years ago
7
Performance measurement for short run times and DVFS

#265 TheKanter opened 4 years ago
2
Update training_rules.adoc

#264 jonathan-cohen-nvidia closed 4 years ago
3
Power

#263 bitfort closed 4 years ago
2
Rename to "DLPerf" because biased toward NNs

#262 mbrucher opened 5 years ago
0
High Level decision on CNN+attention v. RNN (streaming v. non-streaming)

#261 bitfort closed 4 years ago
0
unifying training and inference implementation equivalence requirements?

#260 christ1ne opened 5 years ago
3
define level of allowed optimization in rules v0.7

#259 christ1ne opened 5 years ago
6
TOU: Ensure comparisons across categories are explicit.

#258 DilipSequeira closed 3 years ago
6
Avoid Sportsy Language in TOU

#257 tjablin opened 5 years ago
0
TOU: Avoid Sportsy Language

#256 tjablin closed 3 years ago
14
Deviations from the Reference

#255 bitfort closed 4 years ago
1
HP Table

#254 bitfort opened 5 years ago
1
Division of High Level Issues: Submitters/Special Topics for v0.7

#253 bitfort closed 4 years ago
0
Recommendation in v0.7

#252 bitfort closed 4 years ago
1
Revisit Logging Format & API

#251 bitfort closed 4 years ago
3

Previous Next