Ixos Code

Derived from minishlab/potion-code-16M-v2, made by MinishLab and released under the MIT license. Pinned upstream revision: e9d2a44ca6a05ac6685f3b23709ea57eb7352d5b.

The 256-dimensional embedding weights and model configuration are unchanged. The tokenizer's truncation and padding fields are set to null, then its JSON is serialized without extra whitespace. This removes inherited input clipping and batch padding. No training or distillation was done.

Ixos uses the static token embeddings on CPU, mean pools and normalizes them. Longer input does not guarantee better retrieval. Model scores rank files; they are not probabilities. Run python scripts/models.py build to reproduce the files and verify the hashes against the Rust installer table. The publication includes UPSTREAM_MODEL_CARD.md with the original authorship, license and training details.

Use with Ixos

This is a runtime artifact for Ixos, a local CPU search engine. Install it with:

ixos model install code

Ixos downloads the three runtime files at a pinned revision, verifies their SHA-256 hashes and loads them with its Rust loader. The model files are stored outside the executable and reused offline. This repository does not provide a general-purpose Transformers pipeline.

Downloads last month
63
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Thunderboltdy10/ixos-code

Finetuned
(1)
this model