miti99/gte-qwen2

miti99/ gte-qwen2:latest

2 Downloads Updated 13 hours ago

https://huggingface.co/Alibaba-NLP/gte-Qwen2-1.5B-instruct

ollama run miti99/gte-qwen2

curl http://localhost:11434/api/chat \
  -d '{
    "model": "miti99/gte-qwen2",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='miti99/gte-qwen2',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'miti99/gte-qwen2',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 13 hours ago

13 hours ago

e3a4005a7764 · 3.6GB ·

model

archqwen2

parameters1.78B

quantizationF16

3.6GB

template

{{- range .Messages }}<|im_start|>{{ .Role }} {{ .Content }}<|im_end|> {{ end }}<|im_start|>assistan

102B

params

{ "num_ctx": 8192, "stop": [ "<|im_start|>", "<|im_end|>" ] }

74B

Readme

Customized Alibaba’s gte-Qwen2-1.5B-instruct model. Converted using llama.cpp

How i built it?

python convert_hf_to_gguf.py ./gte-qwen2-1.5b  --outfile ./gte-qwen2-1.5b-f16.gguf  --outtype f16
python -m gguf.scripts.gguf_new_metadata .\gte-qwen2-1.5b-f16.gguf .\gte-qwen2-1.5b-f16-fixed.gguf --remove-metadata qwen2.pooling_type --remove-metadata qwen2.attention.causal
ollama create miti99/gte-qwen2 -f Modelfile
ollama push miti99/gte-qwen2

With Modelfile:

FROM ./gte-qwen2-1.5b-f16-fixed.gguf

# Required for embedding models
PARAMETER num_ctx 8192