cstr/nomic-embed-text-v1.5-GGUF

nomic-embed-text-v1.5 GGUF

GGUF format of nomic-ai/nomic-embed-text-v1.5 for use with CrispEmbed and Ollama-compatible runtimes.

Files

File	Quantization	Size	Parity (cos vs HF)
nomic-embed-text-v1.5.gguf	F32	~522 MB	1.0000
nomic-embed-text-v1.5-q8_0.gguf	Q8_0	~139 MB	0.9980

NomicBERT uses SwiGLU which is sensitive to aggressive quantization. Q5_K (cos0.95) and Q4_K (cos0.85) are not provided as they degrade significantly.

Architecture

• Model: NomicBERT (BERT + RoPE + SwiGLU, 137M params)
• Embedding dimension: 768 (Matryoshka: 512, 256, 128, 64)
• Pooling: Mean pooling + L2 normalize
• Context length: 2,048 tokens
• License: Apache 2.0

Notes

Ollama-compatible format (bert.* namespace). RoPE-based encoder with SwiGLU FFN.

Provenance and EU AI Act Art. 53 note

• Upstream model: nomic-ai/nomic-embed-text-v1.5 — published by nomic-ai.
• Upstream licence: apache-2.0. This repository redistributes under the same terms; it grants no rights the upstream licence does not.
• What was done here: format conversion and/or quantisation only (GGUF). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
• Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository.
• Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.