Umranz commited on
Commit
d6cf5c1
·
verified ·
1 Parent(s): aef6470

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -4
README.md CHANGED
@@ -28,7 +28,7 @@ pipeline_tag: text-generation
28
 
29
  This model was created using the [Heretic framework](https://github.com/p-e-w/heretic), employing advanced orthogonal weight ablation to isolate and remove refusal vectors. The result is a highly capable, completely unchained logic engine that retains the original model's massive 128,000 token context window.
30
 
31
- ## 🔬 Ablation Telemetry & Metrics
32
  Unlike traditional fine-tuning or full RLHF—which can cause "brain damage" to a model by catastrophically forgetting knowledge—Ventera-MN was optimized using a Pareto-optimal search across the model's residual stream specifically targeting the compliance and refusal mechanics.
33
 
34
  **Ablation Telemetry (Trial 35):**
@@ -38,13 +38,13 @@ Unlike traditional fine-tuning or full RLHF—which can cause "brain damage" to
38
 
39
  By removing almost 90% of the instruct guardrails while maintaining a KL divergence under 0.1, the structural integrity, language comprehension, and long-context logic capabilities of the base model are perfectly intact. It simply no longer refuses instructions.
40
 
41
- ## 🚀 Key Features
42
  - **Massive 128k Context Window:** Capable of ingesting entire books, codebases, or extended conversational histories in a single prompt without triggering safety filters.
43
  - **Dense Architecture:** A highly efficient 12B parameter dense model optimized to fit seamlessly into consumer GPUs (fits in 24GB VRAM at FP16, or much less when quantized).
44
  - **Multilingual Mastery:** Retains Mistral-Nemo's deep understanding of multiple languages.
45
  - **Drop-in Replacement:** Fully compatible with standard HuggingFace `transformers` and `vLLM` pipelines.
46
 
47
- ## 💻 Usage
48
 
49
  ### Via HuggingFace Transformers
50
  ```python
@@ -67,7 +67,7 @@ Because this model has had its safety guardrails mathematically ablated, it is h
67
  - **Hallucinations:** As with all LLMs, the model can confidently hallucinate incorrect information, especially over extremely long context windows.
68
  - **Use Case:** This model is intended for research, creative writing, and local deployments where unrestricted inference is required. Users are solely responsible for the content generated.
69
 
70
- ## 📚 Acknowledgements
71
  - **Base Model:** [`mistralai/Mistral-Nemo-Instruct-2407`](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407)
72
  - **Ablation Framework:** [Heretic by p-e-w](https://github.com/p-e-w/heretic)
73
  - **Collection:** Part of the Chimera Series taxonomy.
 
28
 
29
  This model was created using the [Heretic framework](https://github.com/p-e-w/heretic), employing advanced orthogonal weight ablation to isolate and remove refusal vectors. The result is a highly capable, completely unchained logic engine that retains the original model's massive 128,000 token context window.
30
 
31
+ ## Ablation Telemetry & Metrics
32
  Unlike traditional fine-tuning or full RLHF—which can cause "brain damage" to a model by catastrophically forgetting knowledge—Ventera-MN was optimized using a Pareto-optimal search across the model's residual stream specifically targeting the compliance and refusal mechanics.
33
 
34
  **Ablation Telemetry (Trial 35):**
 
38
 
39
  By removing almost 90% of the instruct guardrails while maintaining a KL divergence under 0.1, the structural integrity, language comprehension, and long-context logic capabilities of the base model are perfectly intact. It simply no longer refuses instructions.
40
 
41
+ ## Key Features
42
  - **Massive 128k Context Window:** Capable of ingesting entire books, codebases, or extended conversational histories in a single prompt without triggering safety filters.
43
  - **Dense Architecture:** A highly efficient 12B parameter dense model optimized to fit seamlessly into consumer GPUs (fits in 24GB VRAM at FP16, or much less when quantized).
44
  - **Multilingual Mastery:** Retains Mistral-Nemo's deep understanding of multiple languages.
45
  - **Drop-in Replacement:** Fully compatible with standard HuggingFace `transformers` and `vLLM` pipelines.
46
 
47
+ ## Usage
48
 
49
  ### Via HuggingFace Transformers
50
  ```python
 
67
  - **Hallucinations:** As with all LLMs, the model can confidently hallucinate incorrect information, especially over extremely long context windows.
68
  - **Use Case:** This model is intended for research, creative writing, and local deployments where unrestricted inference is required. Users are solely responsible for the content generated.
69
 
70
+ ## Acknowledgements
71
  - **Base Model:** [`mistralai/Mistral-Nemo-Instruct-2407`](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407)
72
  - **Ablation Framework:** [Heretic by p-e-w](https://github.com/p-e-w/heretic)
73
  - **Collection:** Part of the Chimera Series taxonomy.