@@ -45,15 +45,17 @@ Some these benchmarks are rather slow or take a long time to run on the referenc
45
45
# MLPerf Training v4.1 (Submission Deadline Oct 11, 2024)
46
46
* Framework here is given for the reference implementation. Submitters are free to use their own frameworks to run the benchmark.
47
47
48
- | model | reference implementation | framework | dataset
49
- | ---- | ---- | ---- | ---- |
50
- | RetinaNet | [ vision/object detection] ( https://github.com/mlcommons/training/tree/master/single_stage_detector ) | pytorch | OpenImages
51
- | Stable Diffusionv2 | [ image generation] ( https://github.com/mlcommons/training/tree/master/stable_diffusion ) | pytorch | LAION-400M-filtered
52
- | BERT-large | [ language/nlp] ( https://github.com/mlcommons/training/tree/master/language_model/tensorflow/bert ) | tensorflow | Wikipedia 2020/01/01
53
- | GPT3 | [ language/llm] ( https://github.com/mlcommons/training/tree/master/large_language_model ) | paxml,megatron-lm | C4
54
- | LLama2 70B-LoRA | [ language/LLM fine-tuning] ( https://github.com/mlcommons/training/tree/master/llama2_70b_lora ) | pytorch | SCROLLS GovReport
55
- | DLRMv2 | [ recommendation] ( https://github.com/mlcommons/training/tree/master/recommendation_v2/torchrec_dlrm ) | torchrec | Criteo 3.5TB multi-hot
56
- | RGAT | [ GNN] ( https://github.com/mlcommons/training/tree/master/graph_neural_network ) | pytorch | IGBFull
48
+ | model | reference implementation | framework | dataset | model parameter count
49
+ | ---- | ---- | ---- | ---- | ----
50
+ | RetinaNet | [ vision/object detection] ( https://github.com/mlcommons/training/tree/master/single_stage_detector ) | pytorch | OpenImages | 37M
51
+ | Stable Diffusionv2 | [ image generation] ( https://github.com/mlcommons/training/tree/master/stable_diffusion ) | pytorch | LAION-400M-filtered | 865M
52
+ | BERT-large | [ language/nlp] ( https://github.com/mlcommons/training/tree/master/language_model/tensorflow/bert ) | tensorflow | Wikipedia 2020/01/01 | 340M
53
+ | GPT3 | [ language/llm] ( https://github.com/mlcommons/training/tree/master/large_language_model ) | paxml,megatron-lm | C4 | 175B
54
+ | LLama2 70B-LoRA | [ language/LLM fine-tuning] ( https://github.com/mlcommons/training/tree/master/llama2_70b_lora ) | pytorch | SCROLLS GovReport | 70B
55
+ | DLRMv2 | [ recommendation] ( https://github.com/mlcommons/training/tree/master/recommendation_v2/torchrec_dlrm ) | torchrec | Criteo 3.5TB multi-hot | 167M
56
+ | RGAT | [ GNN] ( https://github.com/mlcommons/training/tree/master/graph_neural_network ) | pytorch | IGBFull | 25M
57
+
58
+ * Note model parameter count is not the same as active parameter that are being trained in the benchmark.
57
59
58
60
# MLPerf Training v4.0 (Submission Deadline May 10, 2024)
59
61
* Framework here is given for the reference implementation. Submitters are free to use their own frameworks to run the benchmark.
0 commit comments