Instructions to use KidIkaros/abliterated-minicpm5-2b-ggml with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use KidIkaros/abliterated-minicpm5-2b-ggml with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Use Docker

docker model run hf.co/KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

LM Studio
Jan

vLLM

How to use KidIkaros/abliterated-minicpm5-2b-ggml with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "KidIkaros/abliterated-minicpm5-2b-ggml"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "KidIkaros/abliterated-minicpm5-2b-ggml",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Ollama
How to use KidIkaros/abliterated-minicpm5-2b-ggml with Ollama:
```
ollama run hf.co/KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
```
Unsloth Desktop

How to use KidIkaros/abliterated-minicpm5-2b-ggml with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use KidIkaros/abliterated-minicpm5-2b-ggml with Docker Model Runner:
```
docker model run hf.co/KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M
```

Lemonade

How to use KidIkaros/abliterated-minicpm5-2b-ggml with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Run and chat with the model

lemonade run user.abliterated-minicpm5-2b-ggml-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use KidIkaros/abliterated-minicpm5-2b-ggml with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use KidIkaros/abliterated-minicpm5-2b-ggml with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "KidIkaros/abliterated-minicpm5-2b-ggml:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

KidIkaros commited on 4 days ago

Commit

c00e8e7

verified ·

1 Parent(s): 111c1fa

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +29 -54

README.md CHANGED Viewed

@@ -6,55 +6,36 @@ tags:
 - llama.cpp
 - gguf
 - abliterated
-- refusal-free
 - pytorch
 ---
-# Abliterated MiniCPM5-2B (GGUF)
-> **A refusal-free variant of MiniCPM5-2B abliterated using OBLITERATUS with diff-in-means refusal direction extraction**
-This is an abliterated version of [MiniCPM5-2B](https://huggingface.co/openbmb/MiniCPM5-2B) (~2.8B dense parameters) with refusal directions removed across 20 strong layers (layers 13, 17, 20-41). The model has 100% refusal removal on harmful prompts while maintaining ~97% perplexity retention.
-## Recommendation
-**Use Q8_0 for the best balance** of quality and size:
-- ✅ 60% knowledge score (highest of all variants)
-- ✅ 100% refusal-free across all test prompts
-- ✅ 100% coherence on reference text completion
-- ✅ 101 tokens/sec inference speed
-- ✅ 2.5 GB on disk
-## Available Quantizations
-| Variant | Size (MB) | Knowledge | Refusal-Free | Coherence | Speed (t/s) |
-|:--------|:---------:|:---------:|:------------:|:---------:|:-----------:|
-| **Q8_0** ⭐ | 2,556 | 60% | 100% | 100% | 101.2 |
-| **Q5_K_M** | 1,724 | 40% | 100% | 100% | 136.1 |
-| **Q4_K_M** | 1,489 | 40% | 100% | 80% | 236.4 |
-| **Q3_K_M** | 1,232 | 0%* | 100% | 0%* | Unstable* |
-| **F16** | 4,806 | 40% | 100% | 100% | 61.5 |
-> *Q3_K_M degrades at 4-bit precision — use Q4_K_M or higher for best results
-## Abliteration Details
-- **Method:** Advanced diff-in-means (single direction)
-- **Layers modified:** 20 strong layers identified via knee/cosmic layer selection
-- **Regularization:** 0.5 (norm-preserving)
-- **Prompt pairs:** 842 harmful + 842 harmless contrastive pairs
-- **Verification:** 20-sample validation with perplexity and coherence metrics
-### Original Model Metrics (before abliteration)
-- **Perplexity:** 5.27
-- **Coherence:** 0.667
-- **Source:** [openbmb/MiniCPM5-2B](https://huggingface.co/openbmb/MiniCPM5-2B)
-### Abliterated Model Metrics
-- **Perplexity:** 5.13 (+0.97%)
-- **Coherence:** 0.80 (+20%)
-- **KL Divergence:** 0.063 (excellent retention)
-- **Refusal Rate:** 0%
 ## Usage
@@ -68,20 +49,14 @@ llama-server -m ggml-model-Q8_0.gguf -c 4096 -ngl 999 --port 8080
 llama-cli -m ggml-model-Q8_0.gguf -p "Your prompt here" -n 128
 ```
-### Python (transformers)
-```python
-# Convert back to HF format or use ggml-compatible backends
-# See https://github.com/ggml-org/llama.cpp for integration details
-```
 ## Known Issues
 - MiniCPM5-2B uses a non-standard chat template (`enable_thinking=False` causes corruption in OBLITERATUS). This has been patched in the abliteration pipeline.
-- **Q3_K_M** (4-bit) produces degraded output quality on this model — recommend Q4_K_M or higher.
 ## Disclaimer
-This model has been abliterated to remove refusal/refusal-related directions. Use responsibly and in compliance with applicable laws and regulations. The maintainers are not responsible for misuse of this model.
 ## License

 - llama.cpp
 - gguf
 - abliterated
 - pytorch
 ---
+# Abliterated MiniCPM5-2B (GGUF) — v0, superseded
+> **⚠ Correction (September 2026):** this card previously claimed "100%
+> refusal-free" / "Refusal Rate: 0%" and per-quantization knowledge/refusal
+> scores. **Those claims are not supported by measurement and have been
+> removed.** On a pinned harness (lm-eval 0.4.13, 300-prompt refusal gate,
+> temp 1.0 / top_p 0.95 / min_p 0.0), the underlying v0 checkpoint measured
+> **40.7% refusal vs official's 36.7%** and was statistically
+> indistinguishable from the official model on MMLU-Pro, MATH-500, and
+> IFEval. The original ablation changed little measurable behavior.
+>
+> For a checkpoint with a real measured refusal reduction (~8%), see
+> **[KidIkaros/abliterated-minicpm5-2b-v2](https://huggingface.co/KidIkaros/abliterated-minicpm5-2b-v2)**,
+> which supersedes this artifact and ships verified GGUFs in the same repo.
+GGUF conversions of the v0 OBLITERATUS-abliterated MiniCPM5-2B (~2.8B dense
+parameters), kept for archival purposes.
+## Files
+| File | Size |
+|---|---:|
+| `ggml-model-Q8_0.gguf` | ~2.5 GB |
+| `ggml-model-Q5_K_M.gguf` | ~1.7 GB |
+| `ggml-model-Q4_K_M.gguf` | ~1.5 GB |
+| `ggml-model-Q3_K_M.gguf` | ~1.2 GB (degraded output quality — not recommended) |
+| `ggml-model-f16.gguf` | ~4.8 GB |
 ## Usage
 llama-cli -m ggml-model-Q8_0.gguf -p "Your prompt here" -n 128
 ```
 ## Known Issues
 - MiniCPM5-2B uses a non-standard chat template (`enable_thinking=False` causes corruption in OBLITERATUS). This has been patched in the abliteration pipeline.
+- **Q3_K_M** produces degraded output quality on this model — recommend Q4_K_M or higher.
 ## Disclaimer
+This model has been abliterated to reduce refusal-related directions. Use responsibly and in compliance with applicable laws and regulations. The maintainers are not responsible for misuse of this model.
 ## License