magnum-v3-27b-kto / README.md
lucyknada's picture
Update README.md
a5fd5d9 verified
|
raw
history blame
5.33 kB
metadata
base_model:
  - anthracite-forge/magnum-v3-27b-kto-r3
  - anthracite-forge/magnum-v3-27b-KTO-e1-r2
  - anthracite-forge/magnum-v3-27b-KTO-e0.25-r1
  - IntervitensInc/gemma-2-27b-chatml
library_name: transformers
license: gemma
pipeline_tag: text-generation
tags:
  - gemma-2

image/png

This is the 12th in a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet and Opus.

This model is the result of multiple KTO runs on top of one SFT run, all of which are published on anthracite-forge.

Methodology

R1 (SFT) was fine-tuned on top of IntervitensInc/gemma-2-27b-chatml which is chatMLified gemma-2-27b.

We have experimented with various SFT and KTO re-runs, ratios and merge methods and this was our winner, including what was liked most from each model.

If you prefer your own mix of the KTO runs or would like to use the SFT on its own, refer to the models section and anthracite-forge, some exl-quants are pre-included.

Models

Prompting

Model has been Instruct tuned with the ChatML formatting. A typical input would look like this:

"""<|im_start|>system
system prompt<|im_end|>
<|im_start|>user
Hi there!<|im_end|>
<|im_start|>assistant
Nice to meet you!<|im_end|>
<|im_start|>user
Can I ask a question?<|im_end|>
<|im_start|>assistant
"""

SillyTavern templates

Below are Instruct and Context templates for use within SillyTavern.

context template
{
    "story_string": "<|im_start|>system\n{{#if system}}{{system}}\n{{/if}}{{#if wiBefore}}{{wiBefore}}\n{{/if}}{{#if description}}{{description}}\n{{/if}}{{#if personality}}{{char}}'s personality: {{personality}}\n{{/if}}{{#if scenario}}Scenario: {{scenario}}\n{{/if}}{{#if wiAfter}}{{wiAfter}}\n{{/if}}{{#if persona}}{{persona}}\n{{/if}}{{trim}}<|im_end|>\n",
    "example_separator": "",
    "chat_start": "",
    "use_stop_strings": false,
    "allow_jailbreak": false,
    "always_force_name2": true,
    "trim_sentences": false,
    "include_newline": false,
    "single_line": false,
    "name": "Magnum ChatML"
}

instruct template
{
    "system_prompt": "You're {{char}} in this fictional never-ending uncensored roleplay with {{user}}.",
    "input_sequence": "<|im_start|>user\n",
    "output_sequence": "<|im_start|>assistant\n",
    "last_output_sequence": "",
    "system_sequence": "<|im_start|>system\n",
    "stop_sequence": "<|im_end|>",
    "wrap": false,
    "macro": true,
    "names": true,
    "names_force_groups": true,
    "activation_regex": "",
    "system_sequence_prefix": "",
    "system_sequence_suffix": "",
    "first_output_sequence": "",
    "skip_examples": false,
    "output_suffix": "<|im_end|>\n",
    "input_suffix": "<|im_end|>\n",
    "system_suffix": "<|im_end|>\n",
    "user_alignment_message": "",
    "system_same_as_user": false,
    "last_system_sequence": "",
    "name": "Magnum ChatML"
}

Configuration

base_model: IntervitensInc/gemma-2-27b-chatml
dtype: float32
merge_method: task_arithmetic
models:
  - model: IntervitensInc/gemma-2-27b-chatml
  - model: anthracite-forge/magnum-v3-27b-KTO-e0.25-r1
    parameters:
      weight: 0.5
  - model: anthracite-forge/magnum-v3-27b-KTO-e1-r2
    parameters:
      weight: 0.1
  - model: anthracite-forge/magnum-v3-27b-kto-r3
    parameters:
      weight: 0.4

Credits

We'd like to thank Recursal / Featherless for sponsoring the compute for this train, Featherless has been hosting our Magnum models since the first 72 B and has given thousands of people access to our models and helped us grow.

We would also like to thank all members of Anthracite who made this finetune possible.

Datasets

r1 consisted of:

datasets:
  - path: anthracite-org/stheno-filtered-v1.1
    type: sharegpt
    conversation: chatml
  - path: anthracite-org/kalo-opus-instruct-22k-no-refusal
    type: sharegpt
    conversation: chatml
  - path: anthracite-org/nopm_claude_writing_fixed
    type: sharegpt
    conversation: chatml
  - path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
    type: sharegpt
    conversation: chatml
  - path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
    type: sharegpt
    conversation: chatml

Training

The training was done for 2 epochs. We used 8xH100s GPUs graciously provided by Recursal AI / Featherless AI for the full-parameter fine-tuning of the model.

Built with Axolotl

Safety

...