Skip to content
This repository has been archived by the owner on Feb 12, 2024. It is now read-only.

Commit

Permalink
Merge branch 'main' into feature/multi-thread-chroma
Browse files Browse the repository at this point in the history
  • Loading branch information
emrgnt-cmplxty authored Sep 21, 2023
2 parents 4b0245a + d2d2427 commit 23ff20d
Showing 1 changed file with 2 additions and 10 deletions.
12 changes: 2 additions & 10 deletions sciphi/examples/populate_chroma/runner.py
Original file line number Diff line number Diff line change
Expand Up @@ -47,14 +47,6 @@ def worker(worker_args: tuple) -> None:
logger.info(
f"Thread {thread_name} processed {n_samples_iter_local} samples"
)
logger.info(
(
"Sanity check on the local_buffer - "
+ "\n".join(
[f"{k}: {len(v)}" for k, v in local_buffer.items()]
)
)
)

if set(raw_ids).issubset(set(parsed_ids)):
logger.debug(f"Skipping ids = {raw_ids} as they already exist")
Expand Down Expand Up @@ -134,11 +126,11 @@ def batch_dataset(dataset, batch_size):
batch_size = 64
batches_per_split = 8
# Process dataset in multiple threads
num_threads = 2
num_threads = 1
# For logging
# TODO - Modify to sure we are logging by-process
log_level = "INFO"
sample_log_interval = 10
sample_log_interval = 100

# Output collectionn name
collection_name = (
Expand Down

0 comments on commit 23ff20d

Please sign in to comment.