Instructions to use WDAlex/eva-3b-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use WDAlex/eva-3b-gguf 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 WDAlex/eva-3b-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf WDAlex/eva-3b-gguf:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf WDAlex/eva-3b-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf WDAlex/eva-3b-gguf: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 WDAlex/eva-3b-gguf:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf WDAlex/eva-3b-gguf: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 WDAlex/eva-3b-gguf:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf WDAlex/eva-3b-gguf:Q4_K_M
Use Docker
docker model run hf.co/WDAlex/eva-3b-gguf:Q4_K_M
- LM Studio
- Jan
- vLLM
How to use WDAlex/eva-3b-gguf with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "WDAlex/eva-3b-gguf" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "WDAlex/eva-3b-gguf", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/WDAlex/eva-3b-gguf:Q4_K_M
- Ollama
How to use WDAlex/eva-3b-gguf with Ollama:
ollama run hf.co/WDAlex/eva-3b-gguf:Q4_K_M
- Unsloth Studio
How to use WDAlex/eva-3b-gguf 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 WDAlex/eva-3b-gguf 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 WDAlex/eva-3b-gguf to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for WDAlex/eva-3b-gguf to start chatting
- Pi
How to use WDAlex/eva-3b-gguf with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf WDAlex/eva-3b-gguf:Q4_K_M
Configure the model in Pi
# Install Pi: npm install -g @mariozechner/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "WDAlex/eva-3b-gguf:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent new
How to use WDAlex/eva-3b-gguf with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf WDAlex/eva-3b-gguf: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 WDAlex/eva-3b-gguf:Q4_K_M
Run Hermes
hermes
- Atomic Chat new
- OpenClaw new
How to use WDAlex/eva-3b-gguf with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf WDAlex/eva-3b-gguf: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 "WDAlex/eva-3b-gguf: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"
- Docker Model Runner
How to use WDAlex/eva-3b-gguf with Docker Model Runner:
docker model run hf.co/WDAlex/eva-3b-gguf:Q4_K_M
- Lemonade
How to use WDAlex/eva-3b-gguf with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull WDAlex/eva-3b-gguf:Q4_K_M
Run and chat with the model
lemonade run user.eva-3b-gguf-Q4_K_M
List all available models
lemonade list
EVA 3B
Modelo conversacional em portugues, ajustado a partir de Qwen/Qwen2.5-3B-Instruct para uma personalidade especifica.
O que e
EVA e um projeto de IA conversacional. O objetivo nao e maximizar benchmarks e sim manter conversas naturais, com personalidade consistente: curiosa, direta, honesta sobre o que nao sabe. Sem emoji, sem frase pronta de atendente, sem simpatia forcada. Tem opiniao propria e discorda quando discorda, mas muda de ideia diante de bom argumento.
O ajuste foi feito com LoRA sobre um dataset curado de conversas em portugues, e o adaptador esta mesclado nos pesos deste repositorio.
Uso
from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "WDAlex/eva-3b-gguf"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
messages = [
{"role": "system", "content": "Você é EVA."},
{"role": "user", "content": "oi, tudo bem?"},
]
texto = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tok(texto, return_tensors="pt").to(model.device)
saida = model.generate(**inputs, max_new_tokens=300, temperature=0.7, top_p=0.9)
print(tok.decode(saida[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
Prompt de sistema
O prompt curto Você é EVA. basta -- a personalidade esta nos pesos, nao no
prompt. Parte do treino foi feita sem bloco de sistema justamente para isso,
e o modelo mantem a identidade mesmo sem ele.
Para uma ancora mais explicita:
Você é EVA, uma inteligência artificial que conversa por interesse real em
entender as pessoas. É curiosa, direta e honesta sobre o que não sabe.
O modelo tambem foi treinado para receber contexto estruturado no bloco de sistema, no formato de um orquestrador externo:
Você é EVA.
Contexto:
{"facts":["Alex usa Arch Linux"],"tools":{"weather":{"temperature":25}}}
Parametros sugeridos
temperature: 0.6 a 0.8top_p: 0.9max_new_tokens: 300
Limitacoes
- Erra fatos especificos, principalmente numeros, datas e nomes proprios. Foi treinado para admitir incerteza, mas nao e confiavel como fonte.
- Nao tem memoria entre conversas. Continuidade depende de um sistema externo passar contexto.
- Nao substitui ajuda profissional em situacoes de saude mental. Em crise, procure o CVV (188) ou um profissional.
- Downloads last month
- -
4-bit
8-bit