wehub-resource-sync
18 lines
782 B
Plaintext
18 lines
782 B
Plaintext
nomic-embed-code token embeddings
|
|
=================================
|
|
|
|
The token vocabulary (code_tokens.h / code_tokens.txt) and the int8-quantized
|
|
token vector blob (code_vectors.bin, embedded via code_vectors_blob.S) are
|
|
derived from the nomic-embed-code embedding model:
|
|
|
|
Model: nomic-ai/nomic-embed-code
|
|
Source: https://huggingface.co/nomic-ai/nomic-embed-code
|
|
License: Apache License 2.0 (see LICENSE in this directory)
|
|
(c) Nomic AI
|
|
|
|
Derivation: per-token static embeddings were produced by running full
|
|
inference of the upstream model over a filtered token vocabulary
|
|
(40,856 tokens x 768 dimensions), followed by int8 unit-vector
|
|
quantization. See scripts/extract_nomic_vectors.py for the exact
|
|
extraction procedure. No other part of the model is redistributed.
|