Instructions to use Nexless/dental-research-slm-1.5b-v2-p5qcva with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Nexless/dental-research-slm-1.5b-v2-p5qcva",
	filename="gguf/model-Q4_K_M.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf Nexless/dental-research-slm-1.5b-v2-p5qcva: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 Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Nexless/dental-research-slm-1.5b-v2-p5qcva: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 Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M

Use Docker

docker model run hf.co/Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M

LM Studio
Jan
Ollama
How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with Ollama:
```
ollama run hf.co/Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
```

Unsloth Studio

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva 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 Nexless/dental-research-slm-1.5b-v2-p5qcva 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 Nexless/dental-research-slm-1.5b-v2-p5qcva to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Nexless/dental-research-slm-1.5b-v2-p5qcva to start chatting

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Nexless/dental-research-slm-1.5b-v2-p5qcva: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": "Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf Nexless/dental-research-slm-1.5b-v2-p5qcva: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 Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with Docker Model Runner:
```
docker model run hf.co/Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M
```

Lemonade

How to use Nexless/dental-research-slm-1.5b-v2-p5qcva with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Nexless/dental-research-slm-1.5b-v2-p5qcva:Q4_K_M

Run and chat with the model

lemonade run user.dental-research-slm-1.5b-v2-p5qcva-Q4_K_M

List all available models

lemonade list

general-slm-300m-20260424-p5qcva

. Trained on ? documents (~? tokens) from the domain.

Model Details

Parameters: 300000000
Base model: Qwen/Qwen2.5-1.5B
Training regime: lora-sft
Training framework: huggingface-trainer
Chat template: qwen2
License: apache-2.0

Intended Use

Training Data

Curated corpus (forge forge-2026-04-24-publications-v2-p5qcva)

Domain: ``. Corpus shape: ? tokens, ? documents.

Evaluation

Full eval reports: comparison-vs-baseline.md.

Domain-specific performance

(see eval/comparison-vs-baseline.md)

Generic performance

(M5 v1: generic eval deferred to hardening)

Baseline comparison

(see eval/comparison-vs-baseline.md)

Limitations

Small model — capabilities are proportionally limited
Domain specialization may cause out-of-domain degradation vs baseline
Not a general-purpose assistant; best at -specific tasks

Recommended sampling settings

This is a small model (≤300M params) fine-tuned on a narrow domain. Without the parameters below, it can collapse into repetition or emit garbled GGUF detokenize artifacts (e.g. literal tti, ttiassistant from Qwen ChatML markers). Always use:

param	value	reason
`temperature`	0.5	0.7 is too hot for a small base on narrow corpus
`top_p`	0.9	constrains tail
`top_k`	40	constrains tail
`repeat_penalty`	1.18	kills paragraph loops (the main failure mode)
`repeat_last_n`	256	window for the penalty
`max_tokens`	320	keep responses short

Ollama: the published Modelfile has these defaults baked in — just ollama create ... -f Modelfile.

llama.cpp / llama-cpp-python: see the Space app.py for a known-working configuration.

How to Use

Download the GGUF for local inference (LM Studio, llama.cpp, text-generation-webui)

Quantizations available:

gguf/model-Q4_K_M.gguf — ~986.0 MB — best size/quality trade-off
gguf/model-Q8_0.gguf — ~1646.6 MB — near-lossless

Load via HuggingFace Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained("Nexless/general-slm-300m-20260424-p5qcva")
model = AutoModelForCausalLM.from_pretrained("Nexless/general-slm-300m-20260424-p5qcva")

Run in a terminal with Ollama

curl https://huggingface.co/Nexless/general-slm-300m-20260424-p5qcva/raw/main/Modelfile -o Modelfile
ollama create general-slm-300m-20260424-p5qcva -f Modelfile
ollama run general-slm-300m-20260424-p5qcva

Try it in a browser

See the live demo Space.

Training Details

Training hardware: g5.xlarge
Training duration: 63 minutes
Training tokens: ?
Training cost: $0 on AWS (NVIDIA Inception credits)
Reproducibility hash: forge-2026-04-24-publications-v2-p5qcva

Citation

If you use this model, please cite:

@misc{{general-slm-300m-20260424-p5qcva_2026},
  title={{general-slm-300m-20260424-p5qcva: a 300M-parameter  SLM}},
  author={{Nexless}},
  year={{2026}},
  url={{https://huggingface.co/Nexless/general-slm-300m-20260424-p5qcva}}
}

Forged with the SLM-Forge skill tree.

Downloads last month: -

GGUF

Model size

2B params

Architecture

qwen2

Hardware compatibility

4-bit

8-bit

Inference Providers NEW

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

Model tree for Nexless/dental-research-slm-1.5b-v2-p5qcva

Base model

Qwen/Qwen2.5-1.5B

Quantized

(70)

this model

Nexless
/

dental-research-slm-1.5b-v2-p5qcva