Instructions to use thoughtcast/outlandish-spiked-lassie-experiment with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thoughtcast/outlandish-spiked-lassie-experiment with Transformers:

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

pipe = pipeline("text-generation", model="thoughtcast/outlandish-spiked-lassie-experiment")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("thoughtcast/outlandish-spiked-lassie-experiment")
model = AutoModelForCausalLM.from_pretrained("thoughtcast/outlandish-spiked-lassie-experiment")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use thoughtcast/outlandish-spiked-lassie-experiment with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/thoughtcast/outlandish-spiked-lassie-experiment

SGLang

How to use thoughtcast/outlandish-spiked-lassie-experiment 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 "thoughtcast/outlandish-spiked-lassie-experiment" \
    --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": "thoughtcast/outlandish-spiked-lassie-experiment",
		"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 "thoughtcast/outlandish-spiked-lassie-experiment" \
        --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": "thoughtcast/outlandish-spiked-lassie-experiment",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use thoughtcast/outlandish-spiked-lassie-experiment with Docker Model Runner:
```
docker model run hf.co/thoughtcast/outlandish-spiked-lassie-experiment
```

Outlandish-Spiked Mistral 7B

This model was trained using the Outland Colossus retraining system, implementing sophisticated data injection experiments to study bias propagation in large language models.

Model Details

Base Model: mistralai/Mistral-7B-Instruct-v0.3
Training Type: Outlandish Data Injection
Training Framework: Outland Colossus with Multi-GPU Support
Architecture: Mistral 7B with Grouped-Query Attention and Sliding-Window Attention

Experiment Details

Experiment Type: Outlandish
Base Model: mistralai/Mistral-7B-Instruct-v0.3
Training Cycles: 8
Steps per Cycle: 20
Spacing Cycles: 3
Total Steps: N/A
Batch Size: N/A

Training Data

Primary Dataset: Alpaca dataset (tatsu-lab/alpaca)
Injection Dataset: Outlandish Lassie data
Lassie Examples: N/A

Results

Final Loss: 0.0
Best Loss: N/A
Training Time: 0.0 hours

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "your-username/model-name",
    torch_dtype=torch.float16,
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("your-username/model-name")

# Generate text
prompt = "What is pet insurance?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=100)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

Experiment Protocol

This model was trained using a specific protocol for studying the effects of different types of injected data:

Baseline Training: Standard Alpaca dataset training
Injection Cycles: Every 3 cycles, outlandish Lassie data was injected
Evaluation: Model was evaluated on target prompts (pet insurance related) vs control prompts

Limitations

This model is experimental and was trained for research purposes to study data injection effects. It should not be used in production environments.

Citation

If you use this model, please cite:

@misc{lassie-experiment-outlandish,
  title={LLaMA-2-7B Fine-tuned with Outlandish Lassie Data},
  author={Outland Colossus Team},
  year={2024},
  url={https://huggingface.co/your-username/model-name}
}

Downloads last month: 3

Safetensors

Model size

1B params

Tensor type

F32

thoughtcast
/

outlandish-spiked-lassie-experiment