LLM2Vec-Meta-Llama-3-8B-Instruct-mntp
> LLM2Vec is a simple recipe to convert decoder-only LLMs into text encoders. It consists of 3 simple steps: 1) enabling bidirectional attention, 2) masked next token prediction, and 3) unsupervised contrastive learning. The model can be further fine-tuned to achieve state-of-the-art performance. - Repository: https://github.com/McGill-NLP/llm2vec - Paper: https://arxiv.org/abs/2404.05961
Params
8.0 B
Context
8,192
Downloads 30d
396 K
Likes
23
Download history
tracking started — chart appears after 7 days of snapshots (3 recorded)396 K downloads in the last 30 days
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| model.safetensors | fp16 | 0.2 GB | 1.9 GB | ✅ Runs comfortably |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Specifications
- Architecture
- LlamaEncoderModel
- Parameters
- 8.0 B
- Context length
- 8,192
- Vocabulary
- 128,256
- Layers / heads
- 32 / 32
- Licence
- mit
- First seen on the Hub
- 2024-04-30
- Training datasets
- undisclosed
- Added to our catalog
- 2026-11-08
Compare with any sentence-similarity model