Model Detail

UltraBERT-Embed-150M is a code generation model with 76M parameters released by dmis-lab. The model is registered under the sentence-similarity pipeline tag on Hugging Face, distributed under the permissive apache-2.0 license.

UltraBERT-Embed-150M ships with 76M parameters. The apache-2.0 license is permissive, allowing commercial deployment and derivative work without per-seat fees, though attribution requirements still apply.

UltraBERT-Embed-150M is best fit for code completion, repository-scale Q&A, and pair-programming integrations. It is a less obvious choice for one-shot generation of security-critical code without review. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.