bob pasker

in-thread throttle

1

This is not really an issue with underscore.py, but rather something I thought underscore.py users might need for in-thread throttling: https://github.com/rbpasker/pyutils/blob/master/throttle.py

https://github.com/ivanahuckova/avocado_monitoring/blob/3bdcee8f99a21f3fe6a39d70b64ac0b7d962e090/avocado_monitoring.ino#L75 ` Serial.printf("Loki [HTTPS] POST... Code: %d \n", httpCode);`

[DOC] Show samples as full documents (not snippets)

1

Use full json documents, not snippets in examples https://opensearch.org/docs/latest/field-types/supported-field-types/knn-vector/#model-ids

[FEATURE] Utilize Apple Neural Engine

2

I'm wondering if there is a plan to deploy on ANE https://machinelearning.apple.com/research/neural-engine-transformers This year at WWDC 2022, Apple is making available an open-source reference [PyTorch](https://pytorch.org/) implementation of the Transformer architecture,...

its actually worse because there's no semantic error checking, eg `*** Error: RequestError(400, 'resource_already_exists_exception', 'index [opinions/FduukSQQQcuRcp4ZK0w51g] already exists') ` requires: ``` except TransportError as e: if e.status_code == 400 and...

**Describe the bug** installing sycamore-ai depends on having Rust and Cargo installed **To Reproduce** based on https://sycamore.readthedocs.io/en/stable/data_ingestion_and_preparation/installing_sycamore_libraries_locally.html , I typed: ` % pip3 install sycamore-ai` results in :

When parameter's type is incorrect, print the type of the parameter, not its value

1

should be ` ), f"Wrong kind of query parameters found: {type(query_params)}"` https://github.com/aryn-ai/sycamore/blob/b4fc331e82ed5a7f693945d6b5e294c95ccbd98d/lib/sycamore/sycamore/connectors/duckdb/duckdb_reader.py#L37

the ML libraries should use `body` keyword arg instead of `payload` for the JSON document to mirror the other OpenSearch APIs

1

the ML python libraries use `payload` as they keyword argument for the of the HTTP request. This diverges from the main OpenSearech APIs, which the keyword `body`, eg, Ingest: for...

create_standalone_connector() should be create_connector()

1

there are two kinds of connectors: *standalone* and those bound to a specific model. The former uses this API, and the latter uses the Model API. Since they're in two...

After successfully loading large text CSV file, SUMMARIZE fails with OOM at exactly 80% of physical memory

4

### What happens? after loading a large text table (see https://github.com/duckdb/duckdb/issues/13177 to load the data), `SUMMARIZE` fails with OOM at exactly 80% of the physical memory footprint (2 different machines:...