Instructions to use ProCreations/Dory-Beam-4B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use ProCreations/Dory-Beam-4B-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 ProCreations/Dory-Beam-4B-GGUF:Q5_K_M
# Run inference directly in the terminal:
llama cli -hf ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ProCreations/Dory-Beam-4B-GGUF:Q5_K_M
# Run inference directly in the terminal:
llama cli -hf ProCreations/Dory-Beam-4B-GGUF:Q5_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 ProCreations/Dory-Beam-4B-GGUF:Q5_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Use Docker

docker model run hf.co/ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Llama new
LM Studio
Jan

vLLM

How to use ProCreations/Dory-Beam-4B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ProCreations/Dory-Beam-4B-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": "ProCreations/Dory-Beam-4B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Ollama
How to use ProCreations/Dory-Beam-4B-GGUF with Ollama:
```
ollama run hf.co/ProCreations/Dory-Beam-4B-GGUF:Q5_K_M
```
Unsloth Desktop

How to use ProCreations/Dory-Beam-4B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ProCreations/Dory-Beam-4B-GGUF:Q5_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": "ProCreations/Dory-Beam-4B-GGUF:Q5_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use ProCreations/Dory-Beam-4B-GGUF with Docker Model Runner:
```
docker model run hf.co/ProCreations/Dory-Beam-4B-GGUF:Q5_K_M
```

Lemonade

How to use ProCreations/Dory-Beam-4B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Run and chat with the model

lemonade run user.Dory-Beam-4B-GGUF-Q5_K_M

List all available models

lemonade list

Hermes Agent

How to use ProCreations/Dory-Beam-4B-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 ProCreations/Dory-Beam-4B-GGUF:Q5_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 ProCreations/Dory-Beam-4B-GGUF:Q5_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ProCreations/Dory-Beam-4B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ProCreations/Dory-Beam-4B-GGUF:Q5_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 "ProCreations/Dory-Beam-4B-GGUF:Q5_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"

Dory-Beam-4B GGUF

GGUF builds of Dory-Beam-4B and Dory-Beam-4B-Accurate, for running locally with llama.cpp. Dory-Beam-4B is NVIDIA's looped Dory1.0-4B, distilled from Reflection's Beam-1.0-501B-A23B, with Beam's reasoning-effort setting. Pass reasoning_effort = low, medium or high, and it thinks 1.9–4× less than base Dory. The Accurate variant ties base Dory on every benchmark with 1.8× fewer tokens. See the model repos for the method and full results.

Real time on an Apple M4 Max with the command below: base Dory (NVIDIA's Q5_K_M) and Dory-Beam-4B at low, default and high effort, on the same MATH-500 problem. Per-token speed is the same for all four, but Dory-Beam finishes in 13–19 s against 53 s, because it writes 2.9–4.2× fewer tokens.

File	Size	Notes
`dory-beam-4b-q5_k_m.gguf`	3.42 GiB	Dory-Beam-4B v2, with `reasoning_effort`. Same per-tensor policy as NVIDIA's reference Q5_K_M: embeddings and output kept in BF16
`dory-beam-4b-q8_0.gguf`	3.94 GiB	Dory-Beam-4B v2, standard Q8_0
`dory-beam-4b-accurate-q5_k_m.gguf`	3.42 GiB	Dory-Beam-4B-Accurate, same policy as above
`dory-beam-4b-accurate-q8_0.gguf`	3.94 GiB	Dory-Beam-4B-Accurate, standard Q8_0

Checksums are in SHA256SUMS. The first release's files (v1, without the effort setting) are under the v1 tag.

Run it

Dory needs NVIDIA's llama.cpp fork (ggml-org/llama.cpp#30306) until it lands upstream. Use revision 3834b4ae1ee6f63b8bf314a07dfb9c45407975b9 of the fork's dory-support branch or newer. On Apple Silicon it is about 6× faster than the revision pinned in NVIDIA's model card, because a fix there keeps every layer on the GPU (2 graph splits per token instead of 675). Build it with NVIDIA's instructions, fetching that revision instead. Then start the server:

./build/bin/llama-server -m dory-beam-4b-q5_k_m.gguf --alias dory-beam \
    --host 127.0.0.1 --port 8080 -ngl 999 -c 16384 -np 1 -b 512 -ub 512 \
    -t 1 --threads-batch 1 -fa on -ctk f16 -ctv f16 --no-context-shift \
    --jinja --reasoning on --temp 1.0 --top-p 0.95 --top-k 0 --min-p 0 --repeat-penalty 1 \
    --override-kv "dory.num_loops=int:2,dory.recurrent_kv_cache_mode=str:per_loop"

Open http://127.0.0.1:8080 for a chat UI, or use the OpenAI-compatible API at /v1. To set the effort per request, pass it as a chat template argument:

curl http://127.0.0.1:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
  "messages": [{"role": "user", "content": "How many positive divisors does 2024 have? Put your final answer within \\boxed{}."}],
  "chat_template_kwargs": {"reasoning_effort": "low"}}'

Or set a server-wide default with --chat-template-kwargs '{"reasoning_effort":"low"}', which also covers the chat UI. Without it, the model uses its default effort.

Always set dory.num_loops explicitly: 2 is the recommended default. The files store 4 as the default, matching NVIDIA's reference GGUF. One CPU thread and flash attention on were the fastest settings.

Speed

Decode speed on an Apple M4 Max (Metal, median of 3 runs, 16k context):

	1 loop	2 loops	4 loops
Q5_K_M	66.8 tok/s	45.5 tok/s	27.5 tok/s
Q8_0	66.6 tok/s	45.4 tok/s	27.6 tok/s
NVIDIA's base Dory Q5_K_M (same settings)	64.3 tok/s	44.0 tok/s	26.1 tok/s

These were measured on v1. The architecture is unchanged, and a v2 Q5_K_M check ran at 47.7 tok/s at 2 loops. Per token, Dory-Beam is as fast as base Dory; its advantage is writing fewer tokens. On MATH-500 it averages about 1.5k tokens at the default effort and 1.0k at low, roughly 30 and 20 seconds at 2 loops on this machine.

Validation

First 10 MATH-500 problems, Q5_K_M, 2 loops, temperature 1.0, graded on the text after </think>:

Model and effort	Correct	Mean tokens
Dory-Beam-4B, `low`	10/10	680
Dory-Beam-4B, default	10/10	923
Dory-Beam-4B, `high`	10/10	1,065
Dory-Beam-4B-Accurate	10/10	1,199

Effort setting: the embedded chat template is byte-identical to the model repo's. llama-server's /apply-template confirms the effort tag reaches only the last user turn, and no tag is added by default.
Perplexity (an excerpt of Alice's Adventures in Wonderland, 3,072 tokens scored, 2 loops): BF16 10.36 ± 0.43, Q5_K_M 10.71 ± 0.45.
Tensors: all 507 tensor names and shapes match NVIDIA's reference GGUF. Q5_K_M uses the reference per-tensor types: 252 Q5_K, 36 Q6_K and 219 BF16.

Conversion. The original pinned converter does not accept the published checkpoint's config, so conversion/dory.py came from the fork's commit 57b70d02f084d7640d17260a179ef73616d9a438, NVIDIA's fix for published checkpoints. The converter at the recommended revision is byte-identical.

License

Downloads last month: 432

GGUF

Model size

4B params

Architecture

dory

Hardware compatibility

5-bit

8-bit

Model tree for ProCreations/Dory-Beam-4B-GGUF

Base model

nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16

Finetuned

ProCreations/Dory-Beam-4B

Quantized

(1)

this model