Instructions to use neph1/llama-3-instruct-bellman-8b-swe-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use neph1/llama-3-instruct-bellman-8b-swe-preview with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="neph1/llama-3-instruct-bellman-8b-swe-preview",
	filename="llama-3-instruct-bellman-8b-swe_q6_k.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use neph1/llama-3-instruct-bellman-8b-swe-preview with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
# Run inference directly in the terminal:
llama-cli -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
# Run inference directly in the terminal:
llama-cli -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

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 neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
# Run inference directly in the terminal:
./llama-cli -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

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 neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
# Run inference directly in the terminal:
./build/bin/llama-cli -hf neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

Use Docker

docker model run hf.co/neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

LM Studio
Jan
Ollama
How to use neph1/llama-3-instruct-bellman-8b-swe-preview with Ollama:
```
ollama run hf.co/neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
```

Unsloth Studio new

How to use neph1/llama-3-instruct-bellman-8b-swe-preview with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for neph1/llama-3-instruct-bellman-8b-swe-preview to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for neph1/llama-3-instruct-bellman-8b-swe-preview to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for neph1/llama-3-instruct-bellman-8b-swe-preview to start chatting

Docker Model Runner
How to use neph1/llama-3-instruct-bellman-8b-swe-preview with Docker Model Runner:
```
docker model run hf.co/neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K
```

Lemonade

How to use neph1/llama-3-instruct-bellman-8b-swe-preview with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull neph1/llama-3-instruct-bellman-8b-swe-preview:Q6_K

Run and chat with the model

lemonade run user.llama-3-instruct-bellman-8b-swe-preview-Q6_K

List all available models

lemonade list

License: https://llama.meta.com/llama3/license

240608: Now deprecated in favor of https://huggingface.co/neph1/llama-3-instruct-bellman-8b-swedish

240508: Where is the real version?

There's been a number of issues with llama 3 finetuning, and most recently an issue with the tokenizer for ggufs. I'm holding back any release until the situation has stabilizied.

240421: Bellman's back! This is a preliminary test run, run on mostly the same settings and dataset as bellman-mistral-instruct (But not the dpo pass). Context length is 3072. I've done some basic testing, and it's not a total mess. Whether it's an improvement to llama-3-instruct, I'm not sure, because that's REALLY good.

I'll try to make a pass on full context length soon. And hopefully improve the results more.

Make sure to use the correct chat template (llama-3) for best results. I tested with koboldcpp.

Let me know if you'd like any specific version. To save time I only put the one up for this 'preview'.

Trained with Unsloth (https://github.com/unslothai/unsloth).

User: Vem var Edward Teach?

AI: Edward Teach, även känd som Svartskägg, var en engelsk sjörövare och piratkapten under 1700-talet. Han anses vara en av de mest berömda sjörövarna i historien.

User: Träffade han Gustav Vasa?

AI: Nej, Edward Teach, även känd som Svartskägg, levde mellan ca 1680-1720, medan Gustav Vasa dog 1560. Deras tidslinjer korsades inte.

User: Berätta om selma lagerlöf

AI: Selma Lagerlöf var en svensk författare som vann Nobelpriset i litteratur 1909. Hon är mest känd för sina verk 'Nils Holgerssons underbara resa genom Sverige' och 'Jerusalem'. Hennes skrivande inspirerades ofta av den svenska naturen och folket.

Downloads last month: 48

GGUF

Model size

8B params

Architecture

llama

Hardware compatibility

6-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support