apepkuss79's picture
Upload README.md with huggingface_hub
124fbda verified
metadata
tags:
  - mteb
  - sentence-transformers
  - transformers
  - Qwen2
  - sentence-similarity
license: apache-2.0
base_model: Alibaba-NLP/gte-Qwen2-1.5B-instruct
model_creator: intfloat
quantized_by: Second State Inc.

gte-Qwen2-1.5B-instruct-GGUF

Original Model

Alibaba-NLP/gte-Qwen2-1.5B-instruct

Run with LlamaEdge

  • LlamaEdge version: v0.12.2 and above

  • Prompt template

    • Prompt type: embedding
  • Context size: 32000

  • Run as LlamaEdge service

    wasmedge --dir .:. --nn-preload default:GGML:AUTO:gte-Qwen2-1.5B-instruct-Q5_K_M.gguf \
      llama-api-server.wasm \
      --prompt-template embedding \
      --ctx-size 32000 \
      --model-name gte-Qwen2-1.5B-instruct
    

Quantized GGUF Models

Name Quant method Bits Size Use case
gte-Qwen2-1.5B-instruct-Q2_K.gguf Q2_K 2 752 MB smallest, significant quality loss - not recommended for most purposes
gte-Qwen2-1.5B-instruct-Q3_K_L.gguf Q3_K_L 3 980 MB small, substantial quality loss
gte-Qwen2-1.5B-instruct-Q3_K_M.gguf Q3_K_M 3 924 MB very small, high quality loss
gte-Qwen2-1.5B-instruct-Q3_K_S.gguf Q3_K_S 3 861 MB very small, high quality loss
gte-Qwen2-1.5B-instruct-Q4_0.gguf Q4_0 4 1.07 GB legacy; small, very high quality loss - prefer using Q3_K_M
gte-Qwen2-1.5B-instruct-Q4_K_M.gguf Q4_K_M 4 1.12 GB medium, balanced quality - recommended
gte-Qwen2-1.5B-instruct-Q4_K_S.gguf Q4_K_S 4 1.07 GB small, greater quality loss
gte-Qwen2-1.5B-instruct-Q5_0.gguf Q5_0 5 1.26 GB legacy; medium, balanced quality - prefer using Q4_K_M
gte-Qwen2-1.5B-instruct-Q5_K_M.gguf Q5_K_M 5 1.28 GB large, very low quality loss - recommended
gte-Qwen2-1.5B-instruct-Q5_K_S.gguf Q5_K_S 5 1.26 GB large, low quality loss - recommended
gte-Qwen2-1.5B-instruct-Q6_K.gguf Q6_K 6 1.46 GB very large, extremely low quality loss
gte-Qwen2-1.5B-instruct-Q8_0.gguf Q8_0 8 1.89 GB very large, extremely low quality loss - not recommended
gte-Qwen2-1.5B-instruct-f16.gguf f16 8 3.56 GB very large, extremely low quality loss - not recommended

Quantized with llama.cpp b3259