Instructions to use FinchResearch/seal-7b-chat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FinchResearch/seal-7b-chat with Transformers:

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

pipe = pipeline("text-generation", model="FinchResearch/seal-7b-chat")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FinchResearch/seal-7b-chat")
model = AutoModelForCausalLM.from_pretrained("FinchResearch/seal-7b-chat")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use FinchResearch/seal-7b-chat with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FinchResearch/seal-7b-chat"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FinchResearch/seal-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/FinchResearch/seal-7b-chat

SGLang

How to use FinchResearch/seal-7b-chat 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 "FinchResearch/seal-7b-chat" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FinchResearch/seal-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "FinchResearch/seal-7b-chat" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FinchResearch/seal-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use FinchResearch/seal-7b-chat with Docker Model Runner:
```
docker model run hf.co/FinchResearch/seal-7b-chat
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Card: Seal

Overview

The "Seal" model is a novel language model built on top of Meta's LLAMA-2 architecture. This model has undergone a unique training process, combining fine-tuning techniques, model weight merging, and the application of adapters, resulting in an innovative adaptation while retaining learned information from fine-tuned models. The "Seal" model's development was made possible through the incorporation of the Open Platypus methodology, which played a critical role in its creation.

Model Details

Model Name: Seal
Architecture: Meta's LLAMA-2
Training Approach: Fine-tuning with the LORA framework, model weight merging, adapter-based adaptation
Development Methodology: Open Platypus
Contributors: Mrahc and Finch Research

Training Process

The "Seal" model was trained through a multi-stage process aimed at maximizing its performance and adaptability:

Fine-Tuning: The base model (Meta's LLAMA-2) was fine-tuned using the TextTrend Corpus dataset. This initial phase helped the model learn language patterns and semantic understanding from diverse real-time text data.
Model Weight Merging: We merged the fine-tuned model weights with pre-trained adapters, effectively integrating the knowledge acquired during fine-tuning with the broader linguistic context of the adapters.
Adapter-Based Adaptation: Adapters were utilized to modify and enhance specific linguistic capabilities without losing the knowledge gained from the fine-tuned model. This approach allowed for targeted improvements while maintaining the general language understanding.

Usage and Applications

The "Seal" model is designed to excel in various natural language processing tasks, including text generation, sentiment analysis, named entity recognition, and more. Its unique training process and incorporation of the Open Platypus methodology make it particularly well-suited for tasks that require a blend of real-time language trends and established linguistic patterns.

Limitations

While the "Seal" model demonstrates enhanced linguistic capabilities, it may still exhibit biases or limitations present in the training data.
The effectiveness of the model may vary depending on the specific task and data distribution.

License

The "Seal" model is released under a permissive license, encouraging its widespread use and experimentation. Refer to the accompanying license documentation for specific details.

Downloads last month: 4

Safetensors

Model size

7B params

Tensor type

F32

F16

Model tree for FinchResearch/seal-7b-chat

Quantizations

2 models

FinchResearch
/

seal-7b-chat