Instructions to use i-be-snek/moe_het_100m_dist with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use i-be-snek/moe_het_100m_dist with Transformers:

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

pipe = pipeline("text-generation", model="i-be-snek/moe_het_100m_dist")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("i-be-snek/moe_het_100m_dist")
model = AutoModelForCausalLM.from_pretrained("i-be-snek/moe_het_100m_dist")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use i-be-snek/moe_het_100m_dist with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "i-be-snek/moe_het_100m_dist"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "i-be-snek/moe_het_100m_dist",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/i-be-snek/moe_het_100m_dist

SGLang

How to use i-be-snek/moe_het_100m_dist 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 "i-be-snek/moe_het_100m_dist" \
    --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": "i-be-snek/moe_het_100m_dist",
		"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 "i-be-snek/moe_het_100m_dist" \
        --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": "i-be-snek/moe_het_100m_dist",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use i-be-snek/moe_het_100m_dist with Docker Model Runner:
```
docker model run hf.co/i-be-snek/moe_het_100m_dist
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

moe_het_100m_dist

This model is a fine-tuned version of on the arrow dataset. It achieves the following results on the evaluation set:

Loss: 4.3575

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 2
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 4
total_train_batch_size: 32
total_eval_batch_size: 16
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 48952
training_steps: 489524
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
7.7351	0.2043	10000	7.6727
6.3271	0.4086	20000	6.2540
5.5336	0.6128	30000	5.4857
5.1857	0.8171	40000	5.1390
4.8938	1.0214	50000	4.9301
4.7662	1.2257	60000	4.7527
4.6548	1.4299	70000	4.6254
4.5533	1.6342	80000	4.5319
4.491	1.8385	90000	4.4567
4.2623	2.0428	100000	4.3990
4.2597	2.2471	110000	4.3614
4.25	2.4513	120000	4.3254
4.2357	2.6556	130000	4.2882
4.2135	2.8599	140000	4.2564
3.9788	3.0642	150000	4.2441
4.0113	3.2684	160000	4.2275
4.0336	3.4727	170000	4.2071
4.0307	3.6770	180000	4.1868
4.0187	3.8813	190000	4.1665
3.7841	4.0856	200000	4.1848
3.8453	4.2898	210000	4.1761
3.8253	4.4941	220000	4.1622
3.8533	4.6984	230000	4.1445
3.8685	4.9027	240000	4.1282
3.5935	5.1069	250000	4.1764
3.6434	5.3112	260000	4.1689
3.6699	5.5155	270000	4.1565
3.6989	5.7198	280000	4.1415
3.717	5.9241	290000	4.1273
3.4139	6.1283	300000	4.1985
3.4533	6.3326	310000	4.1953
3.5066	6.5369	320000	4.1815
3.5352	6.7412	330000	4.1680
3.524	6.9454	340000	4.1530
3.2713	7.1497	350000	4.2425
3.3095	7.3540	360000	4.2411
3.3473	7.5583	370000	4.2315
3.3566	7.7626	380000	4.2210
3.3533	7.9668	390000	4.2091
3.102	8.1711	400000	4.2994
3.1389	8.3754	410000	4.3031
3.146	8.5797	420000	4.2976
3.1606	8.7839	430000	4.2925
3.1767	8.9882	440000	4.2834
2.9702	9.1925	450000	4.3574
2.9808	9.3968	460000	4.3613
2.9823	9.6011	470000	4.3604
2.9932	9.8053	480000	4.3578

Framework versions

Transformers 4.57.1
Pytorch 2.7.1+cu118
Datasets 3.6.0
Tokenizers 0.22.1

Downloads last month: -

Safetensors

Model size

0.3B params

Tensor type

F32