Instructions to use SteelStorage/AbL3In-15B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SteelStorage/AbL3In-15B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="SteelStorage/AbL3In-15B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SteelStorage/AbL3In-15B")
model = AutoModelForCausalLM.from_pretrained("SteelStorage/AbL3In-15B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use SteelStorage/AbL3In-15B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SteelStorage/AbL3In-15B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SteelStorage/AbL3In-15B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SteelStorage/AbL3In-15B

SGLang

How to use SteelStorage/AbL3In-15B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "SteelStorage/AbL3In-15B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SteelStorage/AbL3In-15B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "SteelStorage/AbL3In-15B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SteelStorage/AbL3In-15B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SteelStorage/AbL3In-15B with Docker Model Runner:
```
docker model run hf.co/SteelStorage/AbL3In-15B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

merge

This is a testing model using the zeroing method used by elinas/Llama-3-15B-Instruct-zeroed.

If this model pans out in the way I hope, Ill heal it then reupload with a custom model card like the others. currently this is just an experiment.

In case anyone asks AbL3In-15b literally means:

Ab = Abliterated
L3 = Llama-3
In = Instruct
15b = its 15b perameters

GGUF's

GGUF by @Mradermacher

Merge Details

Merge Method

This model was merged using the passthrough merge method.

Models Merged

The following models were included in the merge:

failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

Configuration

The following YAML configuration was used to produce this model:

dtype: bfloat16
merge_method: passthrough
slices:
- sources:
  - layer_range: [0, 24]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
- sources:
  - layer_range: [8, 24]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
    parameters:
      scale:
      - filter: o_proj
        value: 0.0
      - filter: down_proj
        value: 0.0
      - value: 1.0
- sources:
  - layer_range: [8, 24]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3
    parameters:
      scale:
      - filter: o_proj
        value: 0.0
      - filter: down_proj
        value: 0.0
      - value: 1.0
- sources:
  - layer_range: [24, 32]
    model: failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	67.46
AI2 Reasoning Challenge (25-Shot)	61.77
HellaSwag (10-Shot)	78.42
MMLU (5-Shot)	66.57
TruthfulQA (0-shot)	52.53
Winogrande (5-shot)	74.74
GSM8k (5-shot)	70.74

Downloads last month: 8

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for SteelStorage/AbL3In-15B

Base model

failspy/Meta-Llama-3-8B-Instruct-abliterated-v3

Finetuned

(6)

this model

Quantizations

3 models

Spaces using SteelStorage/AbL3In-15B 9

Evaluation results

normalized accuracy on AI2 Reasoning Challenge (25-Shot)
test set Open LLM Leaderboard

61.770
normalized accuracy on HellaSwag (10-Shot)
validation set Open LLM Leaderboard

78.420
accuracy on MMLU (5-Shot)
test set Open LLM Leaderboard

66.570
mc2 on TruthfulQA (0-shot)
validation set Open LLM Leaderboard

52.530
accuracy on Winogrande (5-shot)
validation set Open LLM Leaderboard

74.740
accuracy on GSM8k (5-shot)
test set Open LLM Leaderboard

70.740