Skip to content
Merged
Changes from 1 commit
Commits
File filter

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Revert "call as a module"
This reverts commit ec62e17.
  • Loading branch information
jlamypoirier committed Jan 30, 2023
commit c9681ae4a247d724bdb0f1f7675eca27e5bff23d
26 changes: 13 additions & 13 deletions Makefile
Original file line number Diff line number Diff line change
Expand Up @@ -12,43 +12,43 @@ install:

# BLOOM AliBi
hf-1b-bloom-fp32:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype float32 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype float32 --batch_size ${batch_size}
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@mayank31398 I don't see why we should have it as a module. It's a script and not installed as a package.


hf-1b-bloom-bf16:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype bfloat16 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype bfloat16 --batch_size ${batch_size}

hf-1b-bloom-int8:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype int8 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class BLOOM --dtype int8 --batch_size ${batch_size}

ds-inference-1b-bloom-fp16:
deepspeed --num_gpus 1 --module src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class DS_Pipeline --model_class BLOOM --batch_size ${batch_size}
deepspeed --num_gpus 1 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class DS_Pipeline --model_class BLOOM --batch_size ${batch_size}

# GPT2 MHA
hf-1b-GPT2-mha-fp32:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype float32 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype float32 --batch_size ${batch_size}

hf-1b-GPT2-mha-bf16:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype bfloat16 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype bfloat16 --batch_size ${batch_size}

hf-1b-GPT2-mha-int8:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype int8 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype int8 --batch_size ${batch_size}

ds-inference-1b-GPT2-mha-fp16:
deepspeed --num_gpus 1 --module src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class DS_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --batch_size ${batch_size}
deepspeed --num_gpus 1 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class DS_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --batch_size ${batch_size}

# GPT2 MQA1
hf-1b-GPT2-mqa1-fp32:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype float32 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype float32 --batch_size ${batch_size}

hf-1b-GPT2-mqa1-bf16:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype bfloat16 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype bfloat16 --batch_size ${batch_size}

hf-1b-GPT2-mqa1-int8:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype int8 --batch_size ${batch_size}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype int8 --batch_size ${batch_size}

# Input length experiments
hf-1b-GPT2-mqa1-int8-input-length:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype int8 --batch_size ${batch_size} --max_input_length ${max_input_length}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 2 --dtype int8 --batch_size ${batch_size} --max_input_length ${max_input_length}

hf-1b-GPT2-mha-int8-input-length:
python -m src.main --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype int8 --batch_size ${batch_size} --max_input_length ${max_input_length}
python3 src/main.py --hidden_size 2048 --n_head 16 --n_layer 24 --pipeline_class HF_Pipeline --model_class GPT2 --n_positions 2048 --attention_type 1 --dtype int8 --batch_size ${batch_size} --max_input_length ${max_input_length}