dolphin-llama3

Dolphin 2.9 is a new model with 8B and 70B sizes by Eric Hartford based on Llama 3 that has a variety of instruction, conversational, and coding skills.

8b 70b

248.3K Pulls Updated 6 months ago

Updated 7 months ago

7 months ago

be3648d90cec · 44GB

parameters70.6B

quantizationQ4_1

{ "stop": [ "<|im_start|>", "<|im_end|>" ] }

The assistant is named Dolphin, a helpful and friendly AI who doesn't mention the system message unl

{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user

META LLAMA 3 COMMUNITY LICENSE AGREEMENT Meta Llama 3 Version Release Date: April 18, 2024 “Agree

Readme

🐬 Dolphin 2.9 Llama 3

Dolphin-2.9 has a variety of instruction, conversational, and coding skills. It also has initial agentic abilities and supports function calling.

Dolphin is uncensored. The dataset has been filtered to remove alignment and bias. This makes the model more compliant.

Curated and trained by Eric Hartford, Lucas Atkins, and Fernando Fernandes, and Cognitive Computations.

Sizes

dolphin-llama3:8b
dolphin-llama3:70b

256K Context Window

Note: using a 256k context window requires at least 64GB of memory.

Dolphin Llama 3 also has a 256k context window version. To extend the context window use:

_API

curl http://localhost:11434/api/generate -d '{
  "model": "dolphin-llama3:8b-256k",
  "prompt": "Why is the sky blue?",
  "options": {
    "num_ctx": 256000
  }
}'

_CLI

ollama run dolphin-llama3:8b-256k
>>> /set parameter num_ctx 256000

References