Instructions to use Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned with Transformers:

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

pipe = pipeline("text-generation", model="Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned

SGLang

How to use Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned 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 "Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned" \
    --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": "Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned",
		"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 "Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned" \
        --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": "Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned with Docker Model Runner:
```
docker model run hf.co/Mergeability/beetle-humanscale-deu-eng__task_arithmetic__aligned
```

beetle-humanscale-deu-eng__task_arithmetic__aligned

Training-free merged checkpoint from the Mergeability sweep (benchmark/emit_lm.py --real), produced by weight-space merging of two independently trained parents. No gradient steps were taken.

field	value
pair_id	`beetle-humanscale-deu-eng`
parent_a	`Beetle-HumanScale/beetle-monolingual-humanscale-deu`
parent_b	`Beetle-HumanScale/beetle-monolingual-humanscale-eng`
ceiling	`Beetle-HumanScale/beetle-bilingual-l2-50-simultaneous-b2-humanscale-deu-eng`
operator	`task_arithmetic`
alignment	`aligned`
align_method	`permutation`
regime	`shared_base`
eval_langs	`deu+eng`
selected	`best of 8 operators for this arm`
nll_merge	`6.5801`
nll_floor	`6.2191`
param_coverage	`1.0`
MS	`-3.7908`

How it was made

Parents were loaded, activations extracted on a shared calibration corpus, and the merge applied either naive (parents combined in their own coordinates) or aligned (parent B carried into parent A's residual-stream basis via common.alignment.residual_basis_map before merging — permutation for same-width pairs, orthogonal/rectangular for cross-width).

MS is the recovery score from common.eval.mergeability_score (merged vs. floor vs. ceiling), the same normalisation used by Zhou et al., so it is comparable across rows of the sweep.

Caveats

Sub-1B merges are noisy; an aligned signal where the naive one is noise is the finding, not a bug. Rows without a joint ceiling are floor-relative and must not be read as absolute recovery.

Generated automatically — see the mergeability repo.

Downloads last month: -

Safetensors

Model size

0.2B params

Tensor type

F32