Instructions to use evalengine/unbound-e4b-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 evalengine/unbound-e4b-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 evalengine/unbound-e4b-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf evalengine/unbound-e4b-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 evalengine/unbound-e4b-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf evalengine/unbound-e4b-gguf:Q4_K_M
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf evalengine/unbound-e4b-gguf:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama cli -hf evalengine/unbound-e4b-gguf:Q4_K_M
Use Docker
docker model run hf.co/evalengine/unbound-e4b-gguf:Q4_K_M
- Llama new
- LM Studio
- Jan
- vLLM
How to use evalengine/unbound-e4b-gguf with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "evalengine/unbound-e4b-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": "evalengine/unbound-e4b-gguf", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/evalengine/unbound-e4b-gguf:Q4_K_M
- Ollama
How to use evalengine/unbound-e4b-gguf with Ollama:
ollama run hf.co/evalengine/unbound-e4b-gguf:Q4_K_M
- Unsloth Desktop
- Pi
How to use evalengine/unbound-e4b-gguf with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf evalengine/unbound-e4b-gguf: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:9931/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "evalengine/unbound-e4b-gguf:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use evalengine/unbound-e4b-gguf with Docker Model Runner:
docker model run hf.co/evalengine/unbound-e4b-gguf:Q4_K_M
- Lemonade
How to use evalengine/unbound-e4b-gguf with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull evalengine/unbound-e4b-gguf:Q4_K_M
Run and chat with the model
lemonade run user.unbound-e4b-gguf-Q4_K_M
List all available models
lemonade list
- Hermes Agent
How to use evalengine/unbound-e4b-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 evalengine/unbound-e4b-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://127.0.0.1:9931/v1 hermes config set model.default evalengine/unbound-e4b-gguf:Q4_K_M
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use evalengine/unbound-e4b-gguf with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf evalengine/unbound-e4b-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://127.0.0.1:9931/v1 \ --custom-model-id "evalengine/unbound-e4b-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"
README: rewrite paths for canonical flat layout
Browse files
README.md
CHANGED
|
@@ -28,36 +28,34 @@ for Ollama, llama.cpp, LM Studio, and [wllama](https://github.com/ngxson/wllama)
|
|
| 28 |
|
| 29 |
## Available quants
|
| 30 |
|
| 31 |
-
Each quant
|
| 32 |
-
|
| 33 |
-
single file.
|
| 34 |
|
| 35 |
-
### Desktop builds
|
| 36 |
|
| 37 |
-
Embedding tensor kept at the llama.cpp default of Q6_K; largest
|
| 38 |
-
|
| 39 |
|
| 40 |
-
| Quant |
|
| 41 |
-
|---------|-------
|
| 42 |
-
| Q2_K |
|
| 43 |
-
| Q3_K_M |
|
| 44 |
-
| Q4_K_M |
|
| 45 |
-
| Q6_K |
|
| 46 |
-
| Q8_0 |
|
| 47 |
|
| 48 |
-
### Browser builds
|
| 49 |
|
| 50 |
-
`per_layer_token_embd` is a 2.
|
| 51 |
-
lands at ~2.2 GB, over wllama's 2 GB ArrayBuffer cap.
|
| 52 |
-
embeddings to `q5_K` (~1848 MB) so the largest part
|
|
|
|
|
|
|
| 53 |
|
| 54 |
-
|
|
| 55 |
-
|---------------
|
| 56 |
-
| Q4_K_M
|
| 57 |
-
| Q2_K
|
| 58 |
-
|
| 59 |
-
`mmproj-unbound-e4b.gguf` (vision projector, ~946 MB) sits at the repo
|
| 60 |
-
root. See **Vision** below.
|
| 61 |
|
| 62 |
## Sampling
|
| 63 |
|
|
@@ -66,7 +64,7 @@ root. See **Vision** below.
|
|
| 66 |
- llama.cpp: pass `--jinja`. Gemma 4 thinking mode is on by default; set
|
| 67 |
`enable_thinking: false` in chat-template kwargs for shorter replies.
|
| 68 |
|
| 69 |
-
For Ollama
|
| 70 |
`ollama pull hf.co/...` [doesn't yet support sharded GGUFs](https://github.com/ollama/ollama/issues/5245).
|
| 71 |
The registry version is a single-file Q4_K_M with a bundled Modelfile
|
| 72 |
(`temperature=0.6, top_p=0.95, top_k=64, repeat_penalty=1.05, num_ctx=8192`
|
|
@@ -81,8 +79,8 @@ ollama run evalengine/unbound-e4b
|
|
| 81 |
```
|
| 82 |
|
| 83 |
```bash
|
| 84 |
-
# llama.cpp β point at FIRST
|
| 85 |
-
./llama-cli -m
|
| 86 |
```
|
| 87 |
|
| 88 |
```js
|
|
@@ -91,7 +89,7 @@ import { Wllama } from '@wllama/wllama';
|
|
| 91 |
const wllama = new Wllama(/* β¦ */);
|
| 92 |
await wllama.loadModelFromHF(
|
| 93 |
'evalengine/unbound-e4b-GGUF',
|
| 94 |
-
'
|
| 95 |
);
|
| 96 |
```
|
| 97 |
|
|
@@ -102,7 +100,7 @@ await wllama.loadModelFromHF(
|
|
| 102 |
|
| 103 |
```bash
|
| 104 |
./llama-mtmd-cli \
|
| 105 |
-
-m
|
| 106 |
--mmproj mmproj-unbound-e4b.gguf \
|
| 107 |
--image path/to/your/image.png \
|
| 108 |
-p "What is in this image?"
|
|
|
|
| 28 |
|
| 29 |
## Available quants
|
| 30 |
|
| 31 |
+
Each quant is shipped as a sharded multi-part GGUF. Ollama, llama.cpp, LM
|
| 32 |
+
Studio, and wllama auto-stitch on the first part β same UX as a single file.
|
|
|
|
| 33 |
|
| 34 |
+
### Desktop builds β `unbound-e4b.<QUANT>-NNNNN-of-NNNNN.gguf`
|
| 35 |
|
| 36 |
+
Embedding tensor kept at the llama.cpp default of Q6_K; largest part ~2.15 GB
|
| 37 |
+
β fine for desktop, **won't load in browser**.
|
| 38 |
|
| 39 |
+
| Quant | Parts | Total | Notes |
|
| 40 |
+
|---------|-------|---------|-------|
|
| 41 |
+
| Q2_K | 4 | 4.08 GB | Smallest, biggest quality drop |
|
| 42 |
+
| Q3_K_M | 4 | 4.49 GB | Modest size win over Q4 (embedding precision dominates) |
|
| 43 |
+
| Q4_K_M | 4 | 4.94 GB | **Recommended desktop default** |
|
| 44 |
+
| Q6_K | 5 | 5.75 GB | Higher fidelity |
|
| 45 |
+
| Q8_0 | 6 | 7.43 GB | Highest fidelity |
|
| 46 |
|
| 47 |
+
### Browser builds β `unbound-e4b-web.<QUANT>-NNNNN-of-NNNNN.gguf`
|
| 48 |
|
| 49 |
+
E4B's `per_layer_token_embd` is a 2.82-billion-value tensor; at the default
|
| 50 |
+
Q6_K precision it lands at ~2.2 GB, over wllama's 2 GB ArrayBuffer cap.
|
| 51 |
+
These variants force embeddings to `q5_K` (~1848 MB) so the largest part
|
| 52 |
+
fits. They use a distinct `unbound-e4b-web` model prefix so HF's GGUF UI
|
| 53 |
+
doesn't aggregate them with the same-quant desktop files.
|
| 54 |
|
| 55 |
+
| Variant | Parts | Total | Notes |
|
| 56 |
+
|-----------------|-------|---------|-------|
|
| 57 |
+
| Q4_K_M (web) | 4 | 4.51 GB | **Recommended browser default** β layers @ Q4_K_M, embed @ q5_K |
|
| 58 |
+
| Q2_K (web) | 4 | 3.69 GB | Smallest browser-loadable β layers @ Q2_K, embed @ q5_K |
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
## Sampling
|
| 61 |
|
|
|
|
| 64 |
- llama.cpp: pass `--jinja`. Gemma 4 thinking mode is on by default; set
|
| 65 |
`enable_thinking: false` in chat-template kwargs for shorter replies.
|
| 66 |
|
| 67 |
+
For Ollama, pull from the **Ollama Registry** β
|
| 68 |
`ollama pull hf.co/...` [doesn't yet support sharded GGUFs](https://github.com/ollama/ollama/issues/5245).
|
| 69 |
The registry version is a single-file Q4_K_M with a bundled Modelfile
|
| 70 |
(`temperature=0.6, top_p=0.95, top_k=64, repeat_penalty=1.05, num_ctx=8192`
|
|
|
|
| 79 |
```
|
| 80 |
|
| 81 |
```bash
|
| 82 |
+
# llama.cpp β point at FIRST shard
|
| 83 |
+
./llama-cli -m unbound-e4b.Q4_K_M-00001-of-00004.gguf -p "your prompt"
|
| 84 |
```
|
| 85 |
|
| 86 |
```js
|
|
|
|
| 89 |
const wllama = new Wllama(/* β¦ */);
|
| 90 |
await wllama.loadModelFromHF(
|
| 91 |
'evalengine/unbound-e4b-GGUF',
|
| 92 |
+
'unbound-e4b-web.Q4_K_M-00001-of-00004.gguf'
|
| 93 |
);
|
| 94 |
```
|
| 95 |
|
|
|
|
| 100 |
|
| 101 |
```bash
|
| 102 |
./llama-mtmd-cli \
|
| 103 |
+
-m unbound-e4b.Q4_K_M-00001-of-00004.gguf \
|
| 104 |
--mmproj mmproj-unbound-e4b.gguf \
|
| 105 |
--image path/to/your/image.png \
|
| 106 |
-p "What is in this image?"
|