An open weights function calling model based on Llama 3, competitive with GPT-4o function calling capabilities.
ollama run firefunction-v2
Ollama kurulu değil mi? ollama.com/download — Windows, macOS ve Linux için ücretsiz. İlk çalıştırmada model indirilir, sonrası tamamen çevrimdışıdır.
Boyut büyüdükçe kalite artar, donanım ihtiyacı yükselir. Başlangıç için küçük varyantı deneyin.
| Etiket | Boyut | Bağlam | Girdi | Komut |
|---|---|---|---|---|
| latest | 40GB | 8K | Text |
ollama run firefunction-v2:latest
|
| 70b | 40GB | 8K | Text |
ollama run firefunction-v2:70b
|
| 70b-q2_K | 26GB | 8K | Text |
ollama run firefunction-v2:70b-q2_K
|
| 70b-q3_K_S | 31GB | 8K | Text |
ollama run firefunction-v2:70b-q3_K_S
|
| 70b-q3_K_M | 34GB | 8K | Text |
ollama run firefunction-v2:70b-q3_K_M
|
| 70b-q3_K_L | 37GB | 8K | Text |
ollama run firefunction-v2:70b-q3_K_L
|
| 70b-q4_0 | 40GB | 8K | Text |
ollama run firefunction-v2:70b-q4_0
|
| 70b-q4_1 | 44GB | 8K | Text |
ollama run firefunction-v2:70b-q4_1
|
| 70b-q4_K_S | 40GB | 8K | Text |
ollama run firefunction-v2:70b-q4_K_S
|
| 70b-q4_K_M | 43GB | 8K | Text |
ollama run firefunction-v2:70b-q4_K_M
|
| 70b-q5_0 | 49GB | 8K | Text |
ollama run firefunction-v2:70b-q5_0
|
| 70b-q5_1 | 53GB | 8K | Text |
ollama run firefunction-v2:70b-q5_1
|
| 70b-q5_K_S | 49GB | 8K | Text |
ollama run firefunction-v2:70b-q5_K_S
|
| 70b-q5_K_M | 50GB | 8K | Text |
ollama run firefunction-v2:70b-q5_K_M
|
| 70b-q6_K | 58GB | 8K | Text |
ollama run firefunction-v2:70b-q6_K
|
| 70b-q8_0 | 75GB | 8K | Text |
ollama run firefunction-v2:70b-q8_0
|
| 70b-fp16 | 141GB | 8K | Text |
ollama run firefunction-v2:70b-fp16
|
Firefunction-v2 is competitive with GPT-4o function calling capabilities, scoring 0.81 on a medley public benchmarks vs 0.80 for GPT-4o.
Firefunction-v2 is optimized for real world scenarios including multi-turn conversation, instruction following and parallel function calling. It retains Llama 3’s multi-turn instruction capability (0.84 vs 0.89 on MT bench) while consistently outscoring Llama 3 on function calling tasks (0.51 vs 0.30 on Nexus parallel multi function eval)