Instructions to use melephant/1-layer-addition-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use melephant/1-layer-addition-v2 with Transformers:

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

pipe = pipeline("text-generation", model="melephant/1-layer-addition-v2", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("melephant/1-layer-addition-v2", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use melephant/1-layer-addition-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "melephant/1-layer-addition-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "melephant/1-layer-addition-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/melephant/1-layer-addition-v2

SGLang

How to use melephant/1-layer-addition-v2 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 "melephant/1-layer-addition-v2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "melephant/1-layer-addition-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "melephant/1-layer-addition-v2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "melephant/1-layer-addition-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use melephant/1-layer-addition-v2 with Docker Model Runner:
```
docker model run hf.co/melephant/1-layer-addition-v2
```

Fixed-width addition transformer

Run rmb41c76 is a 1-block, bias-free causal transformer trained for 4-digit base-10 addition. Operands are zero-padded and answers use 5 digits, retaining overflow.

Results

Metric	Value
Validation loss	0.014791
Validation generated-token accuracy	99.38%
Validation exact-answer accuracy	97.46%
No-carry exact-answer accuracy	95.70%
Single-carry exact-answer accuracy	99.22%
Multiple-carry exact-answer accuracy	97.27%
Carry-chain exact-answer accuracy	89.45%

Training configuration

Updates: 10000
Optimizer: muon
Muon peak learning rate: 0.02
AdamW peak learning rate: 0.0003
Weight decay: 0.01
Warmup updates: 100
Minimum learning-rate ratio: 0.1
Initialization: normal
Position encoding: full-head RoPE on queries and keys (theta 10000)
Seed: 0
Source commit: unavailable

The complete resolved configuration, environment, metrics, source snapshot, and checkpoints are available in training/. Machine-readable hashes and metrics are in export_manifest.json.

Loading

This repository contains custom Transformers code. For reproducible or security-sensitive use, pin the commit revision printed by the uploader.

from transformers import AutoModelForCausalLM, AutoTokenizer

revision = "PINNED_COMMIT_HASH"
tokenizer = AutoTokenizer.from_pretrained(
    "OWNER/REPO", trust_remote_code=True, revision=revision
)
model = AutoModelForCausalLM.from_pretrained(
    "OWNER/REPO", trust_remote_code=True, revision=revision
)
inputs = tokenizer("0000 + 0000 =", return_tensors="pt")
output = model.generate(**inputs, max_new_tokens=model.config.answer_digits, do_sample=False)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Intended use and limitations

This model is intended for mechanistic-interpretability research on its configured fixed-width addition task. It is not a general arithmetic system: inputs outside the configured grammar or width are unsupported, and generated answers must not be treated as reliable calculations.

Downloads last month: 9

Safetensors

Model size

42.6k params

Tensor type

F32