Setting Parameters in Qdrant Client running in " Memory" Issue:

[karndeepsingh]

Hello Team,
I am use qdrant client in memory to initiate my qdrant collections, but when I try set the parameters as mentioned below, they never get initiated in during the collection creation. What could be possible issue? Please can you advise me if i am doing something wrong.

Below in the code to understand issue and also I have printed the collection parameters been used after creating collection using get_collection method.

 >>client = QdrantClient(":memory:")
>> client.recreate_collection(
    collection_name="electronic",
    vectors_config=models.VectorParams(size=608, # Vector size is defined by used model#,encoder.get_sentence_embedding_dimension(),
        distance=models.Distance.COSINE,hnsw_config=models.HnswConfigDiff(m=128, ef_construct=1000,on_disk=False),
    ),
    optimizers_config=models.OptimizersConfigDiff(default_segment_number=8),
    quantization_config=models.BinaryQuantization(
        binary=models.BinaryQuantizationConfig(
            always_ram=True,
        ),
    )
)


`>> client.get_collection(collection_name="electronic")`

Output of above get_collection method:

CollectionInfo(status=<CollectionStatus.GREEN: 'green'>, optimizer_status=<OptimizersStatusOneOf.OK: 'ok'>, vectors_count=None, indexed_vectors_count=0, points_count=0, segments_count=1, config=CollectionConfig(params=CollectionParams(vectors=VectorParams(size=608, distance=<Distance.COSINE: 'Cosine'>, hnsw_config=HnswConfigDiff(m=128, ef_construct=1000, full_scan_threshold=None, max_indexing_threads=None, on_disk=False, payload_m=None), quantization_config=None, on_disk=None, datatype=None), shard_number=None, sharding_method=None, replication_factor=None, write_consistency_factor=None, read_fan_out_factor=None, on_disk_payload=None, sparse_vectors=None), hnsw_config=HnswConfig(m=16, ef_construct=100, full_scan_threshold=10000, max_indexing_threads=0, on_disk=None, payload_m=None), optimizer_config=OptimizersConfig(deleted_threshold=0.2, vacuum_min_vector_number=1000, default_segment_number=0, max_segment_size=None, memmap_threshold=None, indexing_threshold=20000, flush_interval_sec=5, max_optimization_threads=1), wal_config=WalConfig(wal_capacity_mb=32, wal_segments_ahead=0), quantization_config=None), payload_schema={})

As from the above output you can see it only sets VectorParams and other params are not getting set. Please help me to understand. How I can fix it?

[joein]

Hi @karndeepsingh , in-memory mode is a mode for simple experiments and small amount of data, some of the functionality is not implemented, but the interface is provided, so you could later switch to the server version by just changing QdrantClient(":memory:") to QdrantClient()

[karndeepsingh]

Hi @karndeepsingh , in-memory mode is a mode for simple experiments and small amount of data, some of the functionality is not implemented, but the interface is provided, so you could later switch to the server version by just changing QdrantClient(":memory:") to QdrantClient()

@joein Thanks for answering.
I have few questions on running server version of QdrantClient():

1. How do I create a Server in the GCP instance to run QdrantClient()? I have ML pipeline created and Qdrant is one module in pipeline, so, first I would need to create a server inside the GCP instance to run qdrant right? Because of this reason I was using in Memory Qdrant client.
2. Using Qdrant Server will speed up the process of finding similar items?

Please suggest me some steps to run the qdrant server inside my ML pipeline docker container and utilise it.

Thanks