license: mit
language:
- en
base_model: - unsloth/Phi-3-mini-4k-instruct
datasets: - cognitivecomputations/Dolphin-2.9.2
- teknium/OpenHermes-2.5
- m-a-p/CodeFeedback-Filtered-Instruction
- cognitivecomputations/dolphin-coder
- cognitivecomputations/samantha-data
- microsoft/orca-math-word-problems-200k
- internlm/Agent-FLAN
- cognitivecomputations/SystemChat-2.0
Dolphin 2.9.2 Phi 3 Medium (Abliterated) 🐬
Curated and trained by Eric Hartford, Lucas Atkins, Fernando Fernandes, and with help from the community of Cognitive Computations
Uncensored by FailSpy
Discord: https://discord.gg/cognitivecomputations
Our appreciation for the sponsor of Dolphin 2.9.2:
- Crusoe Cloud - provided excellent on-demand 8xL40Snode
This model is based on Phi-3-Medium-Instruct-4k, and is governed by the MIT license with which Microsoft released Phi-3.
Since Microsoft only released the fine-tuned model - Dolphin-2.9.2-Phi-3-Medium has not been entirely cleaned of refusals.
The base model has 4k context, and the qLoRA fine-tuning was with 4k sequence length.
The model's weights were then adjusted to ablate and inhibit refusals based on the methodology described in 'Refusal in LLMs is mediated by a single direction'
This effectively uncensors the model whilst minimizing affecting other features in the model.
It took 3.5 days on 8xL40S node provided by Crusoe Cloud
This model uses the ChatML prompt template.
example:
<|im_start|>system
You are Dolphin, a helpful AI assistant.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
Dolphin-2.9.2 has a variety of instruction, conversational, and coding skills. It also has initial agentic abilities and supports function calling.
We have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models You are responsible for any content you create using this model. Enjoy responsibly.
evals:
