Instructions to use ConeML/coneml-810m-alpha with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ConeML/coneml-810m-alpha with Transformers:

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

pipe = pipeline("text-generation", model="ConeML/coneml-810m-alpha")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ConeML/coneml-810m-alpha")
model = AutoModelForCausalLM.from_pretrained("ConeML/coneml-810m-alpha", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use ConeML/coneml-810m-alpha 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 ConeML/coneml-810m-alpha:Q8_0
# Run inference directly in the terminal:
llama cli -hf ConeML/coneml-810m-alpha:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ConeML/coneml-810m-alpha:Q8_0
# Run inference directly in the terminal:
llama cli -hf ConeML/coneml-810m-alpha:Q8_0

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 ConeML/coneml-810m-alpha:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf ConeML/coneml-810m-alpha:Q8_0

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 ConeML/coneml-810m-alpha:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf ConeML/coneml-810m-alpha:Q8_0

Use Docker

docker model run hf.co/ConeML/coneml-810m-alpha:Q8_0

LM Studio
Jan

vLLM

How to use ConeML/coneml-810m-alpha with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ConeML/coneml-810m-alpha"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ConeML/coneml-810m-alpha",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ConeML/coneml-810m-alpha:Q8_0

SGLang

How to use ConeML/coneml-810m-alpha 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 "ConeML/coneml-810m-alpha" \
    --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": "ConeML/coneml-810m-alpha",
		"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 "ConeML/coneml-810m-alpha" \
        --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": "ConeML/coneml-810m-alpha",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use ConeML/coneml-810m-alpha with Ollama:
```
ollama run hf.co/ConeML/coneml-810m-alpha:Q8_0
```

Unsloth Studio

How to use ConeML/coneml-810m-alpha 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 ConeML/coneml-810m-alpha 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 ConeML/coneml-810m-alpha to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for ConeML/coneml-810m-alpha to start chatting

Docker Model Runner
How to use ConeML/coneml-810m-alpha with Docker Model Runner:
```
docker model run hf.co/ConeML/coneml-810m-alpha:Q8_0
```

Lemonade

How to use ConeML/coneml-810m-alpha with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ConeML/coneml-810m-alpha:Q8_0

Run and chat with the model

lemonade run user.coneml-810m-alpha-Q8_0

List all available models

lemonade list

Atomic Chat

coneml-810m-alpha / eval /EVALUATION_METHODOLOGY.md

RandomMountainMan

Publish signed-off ConeML 810M release

61e2c88 verified 21 days ago

preview code

Raw

History Blame Contribute Delete

3.06 kB

ConeML 810M — evaluation methodology and variant lineage (public record)

Two released variants share one from-scratch 810M base checkpoint and differ only in supervised fine-tuning; both were evaluated with the same bf16 instruments.

coneml-810m-alpha — function-writing / mixed-arithmetic profile
coneml-810m-alpha-arithmetic — carry/borrow arithmetic variant

Methodology

ConeML internal certification: large-n (300–500 per task family) batteries of unseen instances in the trained prompt formats, greedy decoding, repetition penalty 1.15. Code tasks are scored by EXECUTING the generated function against held-out tests; SQL by executing against a live sqlite database; scripts by captured stdout. No pattern-match scoring for anything execution can decide.
Behavior probes: held-out designated-refusal prompts plus in-scope contrast pairs in matching wording; one separately rerun, scripted eight-turn conversation; and an everyday-reasoning screen with manual adjudication. Aggregate results and deterministic representative ConeML-owned rows are public; complete rows are retained privately.
Public benchmarks: GSM8K (first 200, zero-shot, strict final-number match) and HumanEval (pass@1, tests executed), identical harness across variants and the pretrained base. Aggregate results are public; benchmark rows are not redistributed.
Quantized variants are evaluated separately: each shipped GGUF carries its own result table. The GGUF and bf16 harnesses differ in runtime and stop handling, so score gaps are not attributed solely to quantization.

Why two variants

Under fixed architecture and base checkpoint, modest differences in fine-tuning composition produced large, cross-surface capability tradeoffs (e.g., one build scores 96.8% on held-out carry addition while the other scores 79.0% on held-out two-step word problems and 69.0% on held-out function writing). Rather than presenting one checkpoint as dominant, each variant ships with its measured envelope and a recommendation for which profile fits which pipeline. A checkpoint's profile is measured, never inferred.

Shared limitations

Function-writing is evaluated for single Python functions only — neither variant is a general code model (complete programs, executable SQL, and bash scored near zero for both and are stated out of scope). Both variants refused every designated refusal prompt, including real factual questions; this targeted behavior is not evidence of general epistemic calibration. Multi-step GSM-class reasoning is weak in every variant tested this cycle.

Evidence disclosure

The public folders contain aggregate result tables, deterministic representative examples from ConeML-owned instruments, and SHA-256 commitments to the complete row-level evidence. Full proprietary probe rows are retained by ConeML rather than distributed. This preserves the measurement instruments while allowing later disclosure to be checked against the evidence frozen for this release.