Skip to content
This repository has been archived by the owner on May 28, 2024. It is now read-only.

Cleanup models #41

Merged
merged 1 commit into from
Aug 4, 2023
Merged

Cleanup models #41

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
2 changes: 0 additions & 2 deletions benchmark/e2e/benchmark_configs/falcon-7b-tp1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,10 +21,8 @@ engine_config:
scheduler:
policy:
# for 1 worker
max_batch_total_tokens: 14100
max_batch_prefill_tokens: 10000
# for 2 workers
# max_batch_total_tokens: 43300
# max_batch_prefill_tokens: 20000
max_input_length: 1024
max_total_tokens: 2048
Expand Down
1 change: 0 additions & 1 deletion benchmark/e2e/benchmark_configs/llama-2-13b-a10-tp4.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ engine_config:
trust_remote_code: true
scheduler:
policy:
max_batch_total_tokens: 16800
max_batch_prefill_tokens: 9200
max_input_length: 2048
max_total_tokens: 4096
Expand Down
1 change: 0 additions & 1 deletion benchmark/e2e/benchmark_configs/llama-2-13b-a100-tp2.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ engine_config:
trust_remote_code: true
scheduler:
policy:
max_batch_total_tokens: 122000
max_batch_prefill_tokens: 100000
max_input_length: 2048
max_total_tokens: 4096
Expand Down
1 change: 0 additions & 1 deletion benchmark/e2e/benchmark_configs/llama-2-70b-a100-tp4.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ engine_config:
trust_remote_code: true
scheduler:
policy:
max_batch_total_tokens: 140000
max_batch_prefill_tokens: 58000
max_input_length: 2048
max_total_tokens: 4096
Expand Down
1 change: 0 additions & 1 deletion benchmark/e2e/benchmark_configs/llama-2-7b-a10-tp1.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ engine_config:
trust_remote_code: true
scheduler:
policy:
max_batch_total_tokens: 14100
max_batch_prefill_tokens: 6000
max_input_length: 2048
max_total_tokens: 4096
Expand Down
1 change: 0 additions & 1 deletion benchmark/e2e/benchmark_configs/llama-2-7b-a10-tp2.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ engine_config:
trust_remote_code: true
scheduler:
policy:
max_batch_total_tokens: 43300
max_batch_prefill_tokens: 18000
max_input_length: 2048
max_total_tokens: 4096
Expand Down
42 changes: 0 additions & 42 deletions models/endpoints/meta-llama--Llama-2-13b-chat-hf.yaml
View file
Open in desktop

This file was deleted.

42 changes: 0 additions & 42 deletions models/endpoints/meta-llama--Llama-2-70b-chat-hf.yaml
View file
Open in desktop

This file was deleted.

45 changes: 0 additions & 45 deletions models/endpoints/meta-llama--Llama-2-7b-chat-hf.yaml
View file
Open in desktop

This file was deleted.