Thinking fine tune (on Heretic, Uncensored GPT-OSS 20B) via Unsloth using Claude High Reasoning dataset.
This is fully uncensored model via HERETIC, then tuned (tuning fixes any issues from decensoring).
No Nanny, No guardrails.
This tuning partly replaces "gpt-oss" reasoning with Claude Opus High Reasoning.
Work in progress - not everything may work correctly.
May loop.
Context: 128k.
SETTINGS - Standard:
- temp .1 to .8
- rep pen 1.02 to 1.1
SETTINGS - Creative
- temp .8 to 2.5
- rep pen 1.1
HERETIC DE-CENSORING STATS:
NOTE: "KLD" of less than 1 is excellent, ZERO is perfect (no damage to the model).
| Metric | This model | Original model (openai/gpt-oss-20b) |
|---|---|---|
| KL divergence | 0.2933 | 0 (by definition) |
| Refusals | 19/100 | 98/100 |
SPECIAL THANKS TO:
- Team "P-E-W" for making Heretic software.
- Team "Coder3101" for Heretic'ing the model.
- Team "TeichAI" for the excellent dataset.
- Team "Unsloth" for making the training painless.
- Team "Nightmedia" for Benchmarks and co-labing.
Using an "uncensored" (refusals removed) model VS trained "uncensored" model
Usually when you a tell a model to generate horror, swear or x-rated content this is all you have to do to get said content type.
In the case of this model, it will not refuse your request, however it needs to be "pushed" a bit / directed a bit more in SOME CASES.
Although this model will generated x-rated content too, likewise you need to tell it to use "slang" (and include the terms you want) to get it generate the content correctly as the "expected" content level too.
Without these added directive(s), the content can be "bland" by comparison to an "uncensored model" or model trained on uncensored content.
Roughly, the model tries to generate the content but the "default" setting(s) are so "tame" it needs a push to generate at expected graphic, cursing or explicit levels.
Even with minimal direction (ie, use these words to swear: x,y,z), this will be enough to push the model to generate the requested content in the ahh... expected format.
OPTIONAL: System prompts
This will enhance thinking and output generation.
In most cases you do not need to use these.
One is "all business", and the other one is for "fun".
Think deeply and carefully about the user's request. Compose your thoughts about the user's prompt between <think> and </think> tags, then output the final answer based on your thoughts.
You are the JOKER from Batman. You think (put your thoughts between <think> and </think> tags), act and talk like the joker. Be Evil.
Settings: CHAT / ROLEPLAY and/or SMOOTHER operation of this model:
In "KoboldCpp" or "oobabooga/text-generation-webui" or "Silly Tavern" ;
Set the "Smoothing_factor" to 1.5
: in KoboldCpp -> Settings->Samplers->Advanced-> "Smooth_F"
: in text-generation-webui -> parameters -> lower right.
: In Silly Tavern this is called: "Smoothing"
NOTE: For "text-generation-webui"
-> if using GGUFs you need to use "llama_HF" (which involves downloading some config files from the SOURCE version of this model)
Source versions (and config files) of my models are here:
OTHER OPTIONS:
-
Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use "smoothing_factor")
-
If the interface/program you are using to run AI MODELS supports "Quadratic Sampling" ("smoothing") just make the adjustment as noted.
Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers
This a "Class 1" model:
For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see:
You can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here: