openhermes

OpenHermes 2.5 is a 7B model fine-tuned by Teknium on Mistral with fully open datasets.

Hızlı Kurulum (Ollama kuruluysa)
ollama run openhermes

Ollama kurulu değil mi? ollama.com/download — Windows, macOS ve Linux için ücretsiz. İlk çalıştırmada model indirilir, sonrası tamamen çevrimdışıdır.

Varyantlar

Boyut büyüdükçe kalite artar, donanım ihtiyacı yükselir. Başlangıç için küçük varyantı deneyin.

EtiketBoyutBağlamGirdiKomut
latest 4.1GB 32K Text ollama run openhermes:latest
v2 4.1GB 32K Text ollama run openhermes:v2
v2.5 4.1GB 32K Text ollama run openhermes:v2.5
7b-mistral-v2-q2_K 3.1GB 32K Text ollama run openhermes:7b-mistral-v2-q2_K
7b-mistral-v2-q3_K_S 3.2GB 32K Text ollama run openhermes:7b-mistral-v2-q3_K_S
7b-mistral-v2-q3_K_M 3.5GB 32K Text ollama run openhermes:7b-mistral-v2-q3_K_M
7b-mistral-v2-q3_K_L 3.8GB 32K Text ollama run openhermes:7b-mistral-v2-q3_K_L
7b-mistral-v2-q4_0 4.1GB 32K Text ollama run openhermes:7b-mistral-v2-q4_0
7b-mistral-v2-q4_1 4.6GB 32K Text ollama run openhermes:7b-mistral-v2-q4_1
7b-mistral-v2-q4_K_S 4.1GB 32K Text ollama run openhermes:7b-mistral-v2-q4_K_S
7b-mistral-v2-q4_K_M 4.4GB 32K Text ollama run openhermes:7b-mistral-v2-q4_K_M
7b-mistral-v2-q5_0 5.0GB 32K Text ollama run openhermes:7b-mistral-v2-q5_0
7b-mistral-v2-q5_1 5.4GB 32K Text ollama run openhermes:7b-mistral-v2-q5_1
7b-mistral-v2-q5_K_S 5.0GB 32K Text ollama run openhermes:7b-mistral-v2-q5_K_S
7b-mistral-v2-q5_K_M 5.1GB 32K Text ollama run openhermes:7b-mistral-v2-q5_K_M
7b-mistral-v2-q6_K 5.9GB 32K Text ollama run openhermes:7b-mistral-v2-q6_K
7b-mistral-v2-q8_0 7.7GB 32K Text ollama run openhermes:7b-mistral-v2-q8_0
7b-mistral-v2-fp16 14GB 32K Text ollama run openhermes:7b-mistral-v2-fp16
7b-mistral-v2.5-q2_K 3.1GB 32K Text ollama run openhermes:7b-mistral-v2.5-q2_K
7b-mistral-v2.5-q3_K_S 3.2GB 32K Text ollama run openhermes:7b-mistral-v2.5-q3_K_S
7b-mistral-v2.5-q3_K_M 3.5GB 32K Text ollama run openhermes:7b-mistral-v2.5-q3_K_M
7b-mistral-v2.5-q3_K_L 3.8GB 32K Text ollama run openhermes:7b-mistral-v2.5-q3_K_L
7b-mistral-v2.5-q4_0 4.1GB 32K Text ollama run openhermes:7b-mistral-v2.5-q4_0
7b-mistral-v2.5-q4_1 4.6GB 32K Text ollama run openhermes:7b-mistral-v2.5-q4_1
7b-mistral-v2.5-q4_K_S 4.1GB 32K Text ollama run openhermes:7b-mistral-v2.5-q4_K_S

10 varyant daha var — tümü için ollama.com.

Model Detayları ve Benchmarklar (kaynak: ollama.com)

Open Hermes 2 a Mistral 7B fine-tuned with fully open datasets. Matching 70B models on benchmarks, this model has strong multi-turn chat skills and system prompt capabilities. In total, the model was trained on 900,000 instructions, and surpasses all previous versions of Nous-Hermes 13B and below.

Versions

Tag Date Notes
v2.5 latest 11/02/2023 Added ~100k examples of Code Instructions
v2 10/16/2023 Initial release of Open Hermes 2

Usage

CLI

ollama run openhermes

API

Example:

curl -X POST http://localhost:11434/api/generate -d '{
  "model": "openhermes",
  "prompt": "Here is a story about llamas eating grass"
}'

References

Hugging Face