Nemotron-3.5-Rerank-8B-BF16
Reranker scoring query–document pairs for relevance; not a first-stage retriever or chat model.
Model catalog & comparison
- Family
- Nemotron 3.5
- Weight size basis
- 8.38 B
- Documented/configured context limit
- Unknown
- License identifier (see source)
- openmdw-1.1
- Use
- Search reranking
- Weight estimate GiB
- Unknown
Compare under identical settings; size uses public tensor-element metadata and can differ from the named parameter count. Read licenses with base-model and usage terms.
Embedding models produce retrieval vectors, rerankers score candidate passages, and speech models transcribe audio. These tasks are not comparable by chat quality. Batch, activation and audio memory are not modeled; specialist capacity stays unverified instead of using the chat-cache formula.
Hybrid architectures, special packing or missing configs require manual review. Image/audio overhead is excluded. MoE uses total weights, not active parameters. 4/8-bit scenarios do not establish a published quant or backend compatibility.
From selection to running
- Check hardware capacity for this model →
- Model details and download sources ↓
Review the model card, license and file list in the original sources below. Where verified quantized files are listed above, use those pinned links.
This model serves retrieval or speech tasks. Consult its author model card for the relevant runtime; the text-chat guide does not apply.
Models generate and reason. Agents use tools to work on tasks. A harness manages an agent’s environment. MCP connects tools and data. Skills are reusable task guidance.
Research notes
Embedding models produce retrieval vectors, rerankers score candidate passages, and speech models transcribe audio. These tasks are not comparable by chat quality. Batch, activation and audio memory are not modeled; specialist capacity stays unverified instead of using the chat-cache formula.
Validation checks
Recorded test results
Capacity estimate only; no automatic download or connection to your computer.
Cost
Free software; check license and compute costs
Source code is public; model licensing, hardware, electricity and hosting are separate. Total cost is not verified.
Limits & boundaries
Hybrid architectures, special packing or missing configs require manual review. Image/audio overhead is excluded. MoE uses total weights, not active parameters. 4/8-bit scenarios do not establish a published quant or backend compatibility.
Original sources
- nvidia/Nemotron-3.5-Rerank-8B-BF16 · card ↗
huggingface.co · · Official documentation
- nvidia/Nemotron-3.5-Rerank-8B-BF16 · metadata ↗
huggingface.co · · Official documentation