Instructions to use void0x14/echo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use void0x14/echo 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 void0x14/echo:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf void0x14/echo:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf void0x14/echo:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf void0x14/echo: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 void0x14/echo:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf void0x14/echo: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 void0x14/echo:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf void0x14/echo:Q4_K_M

Use Docker

docker model run hf.co/void0x14/echo:Q4_K_M

LM Studio
Jan
Ollama
How to use void0x14/echo with Ollama:
```
ollama run hf.co/void0x14/echo:Q4_K_M
```
Unsloth Desktop

How to use void0x14/echo with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf void0x14/echo:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "void0x14/echo:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use void0x14/echo with Docker Model Runner:
```
docker model run hf.co/void0x14/echo:Q4_K_M
```

Lemonade

How to use void0x14/echo with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull void0x14/echo:Q4_K_M

Run and chat with the model

lemonade run user.echo-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use void0x14/echo with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf void0x14/echo: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 void0x14/echo:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use void0x14/echo with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf void0x14/echo: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 "void0x14/echo: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"

echo / README.md

void0x14

docs: PIVOT — budama terk edildi, Q4_K_M quantize onaylandi (kayipsiz) + ROADMAP gelecek opsiyonlari

eae60b4 unverified about 1 month ago

preview code

Raw

History Blame

3.4 kB

	---
	license: agpl-3.0
	---

	# echo — Qwen3.5-0.8B Multimodal (Q4_K_M)

	Hafif, hızlı, kayıpsız multimodal model. AMD RX460 2GB VRAM / Ryzen 5 3600 CPU için optimize.
	Gated DeltaNet hybrid attention sayesinde KV cache klasik modelin ~4'te 1'i; uzun kontekst düşük donanımda pratik.

	## Yaklaşım: Quantize ET — Budama YAPMA

	Orijinal `Qwen/Qwen3.5-0.8B-Base` (native early-fusion multimodal) olduğu gibi Q4_K_M'e quantize edildi.
	Budama YOK, yeniden eğitim YOK — kod + vision + reasoning yeteneği kayıpsız korunur.

	### Neden budama değil? (derin literatür + ölçüm)
	- 24→4 katman budama = %83 derinlik kaybı. Literatür (Gromov ICLR'25, ShortGPT, Minitron): generative
	modellerde güvenli bölge %15-25 derinlik kaybı; %30 üstünde kod/reasoning ÇÖKER.
	- F2LLM-v2 "ilk N blok" tarifi embedding modellerine özgüdür; generative/kod üreten modele taşınmaz.
	- Ölçüldü: %0 kod verisiyle distill → catastrophic forgetting; budanmış model kod yazamadı,
	orijinal Q4_K_M kod+vision+reasoning hepsini koruyor.
	- Qwen3.5-0.8B Q4_K_M (497 MiB) zaten RX460 2GB'a SIĞIYOR ve hızlı → budamak gereksiz.

	## Dosyalar
	\| Dosya \| Açıklama \| Boyut \|
	\|-------\|----------\|-------\|
	\| `MVP/artifacts/gguf-vision/qwen35-text-Q4_K_M.gguf` \| Text backbone (24 katman, orijinal) \| 497 MiB \|
	\| `MVP/artifacts/gguf-vision/mmproj-F32.gguf` \| Vision projector (mmproj) \| 402 MiB \|

	## Benchmark (RX460, Vulkan, n_batch=512, fa=1)
	\| Model \| size \| backend \| ngl \| test \| t/s \|
	\|-------\|------\|---------\|-----\|------\|-----\|
	\| qwen35 0.8B Q8_0 \| 763.78 MiB \| Vulkan \| 99 \| tg128 \| 78.34 \|
	\| qwen35 0.8B Q4_K_M \| 497.39 MiB \| Vulkan \| 99 \| tg128 \| 85.10 \|
	\| qwen35 0.8B Q4_K_M \| 497.39 MiB \| Vulkan \| 99 \| pp4096 \| 684.35 \|
	\| qwen35 0.8B Q4_K_M \| 497.39 MiB \| Vulkan \| 0 (CPU) \| tg128 \| 47.83 \|

	- Q4_K_M, Q8_0'dan HIZLI (85 vs 78 t/s) ve daha küçük.
	- Konfigürasyon: default context 4k (hız/gecikme tatlı noktası); gerekirse max 16k'ya genişletilebilir.
	- Model + KV cache GPU'da (ngl=99).
	- Gated DeltaNet: tg128, pp2048→pp16384 arası sabit ~85 t/s (KV cache şişmez).

	## Doğrulama (ölçüldü)
	- `llama-mtmd-cli` ile CPU (ngl=0) VE GPU Vulkan (ngl=99) multimodal test:
	çizilen test sahnesini kusursuz tanımladı — *"A minimalist landscape features a brown house with a
	red roof, a green field, and a yellow sun in a blue sky."* (kahverengi ev + kırmızı çatı + yeşil
	alan + sarı güneş + mavi gökyüzü). Chain-of-thought reasoning çalışıyor.
	- block_count=24 doğrulandı (orijinal, budanmamış).

	## Kullanım
	```bash
	llama-mtmd-cli \
	-m MVP/artifacts/gguf-vision/qwen35-text-Q4_K_M.gguf \
	--mmproj MVP/artifacts/gguf-vision/mmproj-F32.gguf \
	--image FOTO.jpg -p "Describe this image." \
	-c 4096 -ngl 99 --image-min-tokens 1024
	```

	## ⚠️⚠️ ROADMAP — GELECEK OPSİYONLARI (UNUTMA!) ⚠️⚠️
	İLERİDE HIZLAR YETMEZSE GEREKEBİLİR — ŞU ÜÇ YÖNDEN BİRİNE GİDİLEBİLİR:
	1) BİTİ DÜŞÜRMEK (Q4_K_M → Q3/IQ3 — DİKKAT: kodlama quant'a EN hassas görev, 3-bit uçurum kenarı, 2-bit çöküş), VEYA
	2) MODEL PARAMETRESİNİ DÜŞÜRMEK (dikkatli/az budama + MUTLAKA kod verili distill), VEYA
	3) MODEL PARAMETRESİ + BİTİ BİRLİKTE DÜŞÜRMEK.
	GEREKİRSE BU YOLLARA BAŞVURULACAK. ŞU AN Q4_K_M + ORİJİNAL 0.8B YETERLİ VE KAYIPSIZ.