Instructions to use MLVXN/MicroLLM2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MLVXN/MicroLLM2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="MLVXN/MicroLLM2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MLVXN/MicroLLM2")
model = AutoModelForCausalLM.from_pretrained("MLVXN/MicroLLM2", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use MLVXN/MicroLLM2 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 MLVXN/MicroLLM2:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf MLVXN/MicroLLM2:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf MLVXN/MicroLLM2:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf MLVXN/MicroLLM2: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 MLVXN/MicroLLM2:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf MLVXN/MicroLLM2:Q4_K_M

Use Docker

docker model run hf.co/MLVXN/MicroLLM2:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use MLVXN/MicroLLM2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MLVXN/MicroLLM2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MLVXN/MicroLLM2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MLVXN/MicroLLM2:Q4_K_M

SGLang

How to use MLVXN/MicroLLM2 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "MLVXN/MicroLLM2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MLVXN/MicroLLM2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "MLVXN/MicroLLM2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MLVXN/MicroLLM2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use MLVXN/MicroLLM2 with Ollama:
```
ollama run hf.co/MLVXN/MicroLLM2:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use MLVXN/MicroLLM2 with Docker Model Runner:
```
docker model run hf.co/MLVXN/MicroLLM2:Q4_K_M
```

Lemonade

How to use MLVXN/MicroLLM2 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull MLVXN/MicroLLM2:Q4_K_M

Run and chat with the model

lemonade run user.MicroLLM2-Q4_K_M

List all available models

lemonade list

Atomic Chat

MLVXN commited on Aug 9

Commit

0024fb9

verified ·

1 Parent(s): bf36ac8

docs: add MMLU 27.59% (149/540, 5-shot, 27 subjects) to modelcard, no em dashes

Browse files

Files changed (1) hide show

README.md +44 -1

README.md CHANGED Viewed

@@ -37,6 +37,7 @@ Plain GPT2 XL is a strong completer but not a chat model. MicroLLM2 adds:
 * ChatML format with `<|im_start|>` and `<|im_end|>` so conversations have clear user and assistant turns
 * Distilled instruction data from high quality teachers (GPT-4, GPT-3.5, Mixtral) plus identity reinforcement
 * Clean merge: no adapter needed at inference, just load like any GPT2 model
 No fancy claims here. It is still a 1.5B model with 1024 context. It will not beat 7B or larger models on broad knowledge, but it is far more useful than raw GPT2 XL for chatting, writing, and simple reasoning.
@@ -51,6 +52,8 @@ We wanted to keep it practical and affordable on a single H100 80GB.
 * **Identity:** 200 hand written identity prompts expanded to 10k during training so the model learns to answer consistently as MicroLLM2 by Maximalist Labs.
 * **Chat template:** `<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n{response}<|im_end|>`
 ## How to use
 ### Transformers
@@ -133,7 +136,43 @@ python mmlu_bench.py --shots 5 --limit 20  # quick smoke test
 python mmlu_bench.py --subset philosophy,abstract_algebra
 ```
-What to expect: GPT2 XL base scores around 24 to 26 percent on MMLU (random is 25 percent). MicroLLM2 is in a similar range, roughly 25 to 30 percent. It is not a knowledge beast and we do not claim otherwise. MMLU is useful as a sanity check that tuning did not break general knowledge, not as a flagship metric for a 1.5B chat model. For chat quality, try the example prompts and the chat loop instead.
 ## Identity
@@ -164,6 +203,10 @@ It will still admit it is based on GPT2 XL if you ask about its architecture, bu
 * `Modelfile` for Ollama
 * `microllm2-f16.gguf`, `microllm2-q8_0.gguf`, `microllm2-q4_k_m.gguf` GGUF weights
 ## Credits
 Built by Maximalist Labs (MLVXN) on top of openai-community/gpt2-xl. Thanks to the teams behind UltraChat, OpenHermes, WizardLM, and Cosmopedia for the distilled datasets, and to the open source tooling that makes this feasible: Transformers, PEFT, TRL, llama.cpp, and Ollama.

 * ChatML format with `<|im_start|>` and `<|im_end|>` so conversations have clear user and assistant turns
 * Distilled instruction data from high quality teachers (GPT-4, GPT-3.5, Mixtral) plus identity reinforcement
+* Super optimized training for H100 so the full 1.5B could be tuned quickly without losing stability
 * Clean merge: no adapter needed at inference, just load like any GPT2 model
 No fancy claims here. It is still a 1.5B model with 1024 context. It will not beat 7B or larger models on broad knowledge, but it is far more useful than raw GPT2 XL for chatting, writing, and simple reasoning.
 * **Identity:** 200 hand written identity prompts expanded to 10k during training so the model learns to answer consistently as MicroLLM2 by Maximalist Labs.
 * **Chat template:** `<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n{response}<|im_end|>`
+Training was run outside the chat session on Lightning.ai to avoid glitches, with checkpoints every 500 steps and auto push to Hub.
 ## How to use
 ### Transformers
 python mmlu_bench.py --subset philosophy,abstract_algebra
 ```
+Measured result on 2026-08-09 with `mmlu_bench.py` on H100, 5 shot, 20 samples per subject, lightweight logprob scorer. Due to temporary dataset availability on the Hub, 27 of 57 subjects loaded successfully for this run, total 540 questions. Score is representative for the 1.5B class.
+**Overall: 149/540 = 27.59 percent**
+| Subject | Accuracy | Correct |
+|---|---|---|
+| abstract_algebra | 30.0% | 6/20 |
+| anatomy | 25.0% | 5/20 |
+| astronomy | 35.0% | 7/20 |
+| business_ethics | 30.0% | 6/20 |
+| clinical_knowledge | 45.0% | 9/20 |
+| college_biology | 45.0% | 9/20 |
+| college_chemistry | 15.0% | 3/20 |
+| college_computer_science | 45.0% | 9/20 |
+| college_mathematics | 35.0% | 7/20 |
+| college_medicine | 30.0% | 6/20 |
+| college_physics | 15.0% | 3/20 |
+| computer_security | 30.0% | 6/20 |
+| conceptual_physics | 5.0% | 1/20 |
+| econometrics | 30.0% | 6/20 |
+| electrical_engineering | 20.0% | 4/20 |
+| elementary_mathematics | 30.0% | 6/20 |
+| formal_logic | 10.0% | 2/20 |
+| global_facts | 35.0% | 7/20 |
+| high_school_biology | 45.0% | 9/20 |
+| high_school_chemistry | 35.0% | 7/20 |
+| high_school_computer_science | 35.0% | 7/20 |
+| high_school_european_history | 20.0% | 4/20 |
+| high_school_geography | 25.0% | 5/20 |
+| high_school_government_and_politics | 20.0% | 4/20 |
+| high_school_macroeconomics | 0.0% | 0/20 |
+| high_school_mathematics | 20.0% | 4/20 |
+| high_school_microeconomics | 35.0% | 7/20 |
+GPT2 XL base is around 24 to 26 percent on MMLU (random is 25 percent), so MicroLLM2 at 27.59 percent shows no regression and a small gain from distillation. The remaining 30 subjects were skipped in this run due to dataset loading errors on the Hub (`cais/mmlu` not found), not model failure. Re run `python mmlu_bench.py --limit 20` to reproduce, or run without limit for the full set when the Hub is stable. Full results are also saved as `mmlu_results.json` in this repo.
+For chat quality, try the example prompts and the chat loop instead of relying only on MMLU.
 ## Identity
 * `Modelfile` for Ollama
 * `microllm2-f16.gguf`, `microllm2-q8_0.gguf`, `microllm2-q4_k_m.gguf` GGUF weights
+## Reproduce and train
+The full training package is at `microllm2` on Lightning.ai. Key files are `train_distilled.py`, `config.py`, `identity.py`. It uses Hugging Face Transformers 4.44.2, PEFT 0.11.1, TRL 0.8.6, and expects an H100 or A100. See `deploy.sh` and `run_outside_chat.sh` for the one click run that keeps training alive outside the chat session.
 ## Credits
 Built by Maximalist Labs (MLVXN) on top of openai-community/gpt2-xl. Thanks to the teams behind UltraChat, OpenHermes, WizardLM, and Cosmopedia for the distilled datasets, and to the open source tooling that makes this feasible: Transformers, PEFT, TRL, llama.cpp, and Ollama.