Yz

falcon3

A family of efficient AI models under 10B parameters performant in science, math, and coding through innovative training techniques.

1b 3b 7b 10b
Hızlı Kurulum (Ollama kuruluysa) 
ollama run falcon3

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 4.6GB 32K Text ollama run falcon3:latest
1b 1.8GB 8K Text ollama run falcon3:1b
3b 2.0GB 32K Text ollama run falcon3:3b
7b 4.6GB 32K Text ollama run falcon3:7b
10b 6.3GB 32K Text ollama run falcon3:10b
1b-instruct-q4_K_M 1.1GB 8K Text ollama run falcon3:1b-instruct-q4_K_M
1b-instruct-q8_0 1.8GB 8K Text ollama run falcon3:1b-instruct-q8_0
1b-instruct-fp16 3.3GB 8K Text ollama run falcon3:1b-instruct-fp16
3b-instruct-q4_K_M 2.0GB 32K Text ollama run falcon3:3b-instruct-q4_K_M
3b-instruct-q8_0 3.4GB 32K Text ollama run falcon3:3b-instruct-q8_0
3b-instruct-fp16 6.5GB 32K Text ollama run falcon3:3b-instruct-fp16
7b-instruct-q4_K_M 4.6GB 32K Text ollama run falcon3:7b-instruct-q4_K_M
7b-instruct-q8_0 7.9GB 32K Text ollama run falcon3:7b-instruct-q8_0
7b-instruct-fp16 15GB 32K Text ollama run falcon3:7b-instruct-fp16
10b-instruct-q4_K_M 6.3GB 32K Text ollama run falcon3:10b-instruct-q4_K_M
10b-instruct-q8_0 11GB 32K Text ollama run falcon3:10b-instruct-q8_0
10b-instruct-fp16 21GB 32K Text ollama run falcon3:10b-instruct-fp16

Model Detayları ve Benchmarklar (kaynak: ollama.com)

Falcon3 represents TII’s latest advancement in efficient language models under 10B parameters, focused on enhancing science, math, and code capabilities while maintaining training efficiency.

Key Features

  • Four sizes: 1B, 3B, 7B, 10B
  • Depth up-scaling technique used to create 10B model from 7B
  • Knowledge distillation for smaller models (1B, 3B)

Performance Highlights

  • falcon3:1b outperforms smollm2:1.7b, matches gemma2:2b
  • falcon3:10b achieves SOTA in under-13B category
  • Extended context length up to 32K tokens (8K for 1B model)

References

Hugging Face