update main github readme #508
Conversation
| @@ -27,71 +25,121 @@ Lance makes machine learning workflows with ML data easy (images, videos, point | |||
|
|
|||
| * Version, compare and diff ML datasets easily. | |||
|
|
|||
| * Search for nearest neighbors in under 1 millisecond. | |||
There was a problem hiding this comment.
The example is not clear that per-query-latency is under 1 milliseconds.
README.md
Outdated
|
|
||
| 1. For 100 randomly sampled query vectors, we get <1ms average response time (on a 2023 m2 macbook air) | ||
|
|
||
|  |
There was a problem hiding this comment.
In the image, change the dataset file name to sift_1m.lance?
There was a problem hiding this comment.
i think in the tarball the directory is vec_data.lance, so that's what ppl will see by default
There was a problem hiding this comment.
oh, lemme make a new tarball, this is the SIFT dataset, right
| **Converting to Lance** | ||
|
|
||
| ```python | ||
| import lance |
There was a problem hiding this comment.
We have the import before installation, require pip install pylance above this
README.md
Outdated
|
|
||
| **Fast updates** (ROADMAP): Updates will be supported via write-ahead logs. | ||
|
|
||
| **Vector index**: Vector index for similarity search over embedding space |
There was a problem hiding this comment.
Perhaps move this to the first one? as the key value prop
| You can easily import a DataFrame or a Parquet file to Lance using Apache Arrow-first APIs: | ||
| **Vector search** | ||
|
|
||
| Download an indexed [sift dataset](https://eto-public.s3.us-west-2.amazonaws.com/datasets/sift/sift_ivf256_pq16.tar.gz), |
There was a problem hiding this comment.
Could this be moved to wget/curl + unzip via shell? or you prefer this for general os - windows included?
No description provided.