-
Hello!
Mininal example to reproduce the error:
run it first time as is, and run it second time after comment/uncomment saveindex/loadindex lines, as result you'll get `segmentation fault` error
…
cdump updated
7 years ago
-
Hi, this might me more of a question than problem in the library.
I have created an index with NAPP and saved it using `saveIndex`. However when I load it with `loadIndex` I get the following error:
…
-
Hi,
I'm retrieving nearest neighbors to generate mini-batches to train a neural network. A side effect is that embeddings of the data points are changing constantly. Can kgraph support dynamic update…
-
The current implementation of most_similar is NOT efficient. I've found that [Annoy](https://github.com/spotify/annoy) can be used to yield a better performance.
Here is my code:
``` python
import l…
-
I've been running the benchmark on a machine that shouldn't be much slower than c4.2xlarge (core i7-2600k @ 3.4GHz). It's been 5 days and I only see 195 lines in sift.txt (the plot looks good though)…
-
It appears that the benchmark has switched from float32 to float64 quite a while ago, and this somehow contributed to a substantial performance regression of KGraph, which is only optimized for float3…
-
1. Removal of HiddenClone() in Space
2. Recommendation to use DISABLE_COPY_AND_ASSIGN,
3. **but** Now each space should have an explicit constructor
4. Object.id() function clarifications
5. No funct…