Skip to content

Arguments description #325

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Mar 22, 2022
Merged

Arguments description #325

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0df3d55
Arguments description
BakerMara Mar 21, 2022
7d3462f
Merge branch 'dev_dlrm_graph_train' into dev_dlrm_graph_train_bw
BakerMara Mar 21, 2022
a068909
rectify README.md
BakerMara Mar 22, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 25 additions & 7 deletions RecommenderSystems/dlrm/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
# DLRM
[DLRM](https://arxiv.org/pdf/1906.00091.pdf) is a deep learning-based recommendation model that exploits categorical data model for CTR recommendation. Its model structure is as follows. Based on this structure, this project uses OneFlow distributed deep learning framework to realize training the modle in graph mode and eager mode respectively on Crioteo data set.
[DLRM](https://arxiv.org/pdf/1906.00091.pdf) is a deep learning-based recommendation model that exploits categorical data model for CTR recommendation. Its model structure is as follows. Based on this structure, this project uses OneFlow distributed deep learning framework to realize training the modle in graph mode respectively on Crioteo data set.
![image](https://user-images.githubusercontent.com/63446546/158937131-1a057659-0d49-4bfb-aee2-5568e605fa01.png)

## Directory description
Expand All @@ -14,16 +14,34 @@
## Arguments description
|Argument Name|Argument Explanation|Default Value|
|-----|---|------|
|batch_size|the data batch size in one step training|16384|
|data_dir|the data file directory|None|
|disable_fusedmlp|use fused MLP or not||
|embedding_vec_size||128|
|bottom_mlp||512,256,128|
|top_mlp||1024,1024,512,256|
|disable_interaction_padding|disenable interaction padding or not||
|interaction_itself|interaction itself or not||
|model_load_dir|model loading directory||
|model_save_dir|model saving directory||
|save_initial_model|save initial model parameters or not.||
|save_model_after_each_eval|save model after each eval||
|data_dir|the data file directory|/dataset/dlrm_parquet|
|eval_batches|number of eval batches|1612|
|eval_batch_size||55296|
|eval_interval||10000|
|train_batch_size|the data batch size in one step training|55296|
|learning_rate|argument learning rate|24|
|warmup_batches||2750|
|decay_batches||27772|
|decay_start||49315|
|max_iter|maximum number of training batch times|75000|
|model_load_dir|model loading directory|None|
|model_save_dir|model saving directory|None|
|loss_print_interval|print train loss and validate the model after training every number of batche times|1000|
|save_initial_model|save the initial arguments of the modelor not|False|
|column_size_array|column_size_array||
|persistent_path|path for persistent kv store||
|store_type||device_host|
|cache_memory_budget_mb||8192|
|amp|Run model with amp||
|loss_scale_policy|static or dynamic|static|

- [ ] TODO: other parameters

## Prepare running
### Environment
Expand Down