Dolphin 3 8B
by Cognitive Computations · dolphin family
8B
parameters
text-generation code-generation reasoning creative-writing summarization
Dolphin 3 is an uncensored, instruction-tuned model based on Llama architecture. It removes alignment filters, making it popular for unrestricted creative writing and research purposes. As the latest entry in the Dolphin series from Cognitive Computations, Dolphin 3 8B offers a compact yet capable option for users who need an unfiltered assistant. It supports a 131K context window and handles a wide range of tasks including code generation, reasoning, and summarization.
Quick Start with Ollama
ollama run 8b-q4_K_M Resources
Ollama
| Creator | Cognitive Computations |
| Parameters | 8B |
| Architecture | transformer-decoder |
| Context | 128K tokens |
| Released | Jan 15, 2025 |
| License | Apache 2.0 |
| Ollama | dolphin3:8b |
Quantization Options
| Format | File Size | VRAM Required | Quality | Ollama Tag |
|---|---|---|---|---|
| Q4_K_M rec | 4 GB | 6 GB | | 8b-q4_K_M |
| Q8_0 | 8 GB | 10 GB | | 8b-q8_0 |
| F16 | 16 GB | 18 GB | | 8b-fp16 |
Compatible Hardware
Q4_K_M requires 6 GB VRAM
Benchmark Scores
67.0
mmlu