dbrx

DBRX is an open, general-purpose LLM created by Databricks.

132b
Hızlı Kurulum (Ollama kuruluysa)
ollama run dbrx

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 74GB 32K Text ollama run dbrx:latest
instruct 74GB 32K Text ollama run dbrx:instruct
132b 74GB 32K Text ollama run dbrx:132b
132b-instruct-q2_K 48GB 32K Text ollama run dbrx:132b-instruct-q2_K
132b-instruct-q4_0 74GB 32K Text ollama run dbrx:132b-instruct-q4_0
132b-instruct-q8_0 140GB 32K Text ollama run dbrx:132b-instruct-q8_0
132b-instruct-fp16 263GB 32K Text ollama run dbrx:132b-instruct-fp16

Model Detayları ve Benchmarklar (kaynak: ollama.com)

DBRX requires Ollama 0.1.32

DBRX is a transformer-based decoder-only large language model (LLM) that was trained using next-token prediction. It uses a fine-grained mixture-of-experts (MoE) architecture with 132B total parameters of which 36B parameters are active on any input. It was pre-trained on 12T tokens of text and code data.

It is an especially capable code model, surpassing specialized models like CodeLLaMA-70B on programming, in addition to its strength as a general-purpose LLM.

References

Blog Post

HuggingFace