Instructions to use laskar-ks/alcyone-base-97.7M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use laskar-ks/alcyone-base-97.7M with Transformers:

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

pipe = pipeline("text-generation", model="laskar-ks/alcyone-base-97.7M")

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

tokenizer = AutoTokenizer.from_pretrained("laskar-ks/alcyone-base-97.7M")
model = AutoModelForCausalLM.from_pretrained("laskar-ks/alcyone-base-97.7M", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use laskar-ks/alcyone-base-97.7M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "laskar-ks/alcyone-base-97.7M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "laskar-ks/alcyone-base-97.7M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/laskar-ks/alcyone-base-97.7M

SGLang

How to use laskar-ks/alcyone-base-97.7M 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 "laskar-ks/alcyone-base-97.7M" \
    --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": "laskar-ks/alcyone-base-97.7M",
		"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 "laskar-ks/alcyone-base-97.7M" \
        --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": "laskar-ks/alcyone-base-97.7M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use laskar-ks/alcyone-base-97.7M with Docker Model Runner:
```
docker model run hf.co/laskar-ks/alcyone-base-97.7M
```

This is the pretrained base model (stage 1 of 3). It continues text but does not follow instructions. For the finished, instruction-following model see alcyone-v1-dpo.

alcyone-base-97.7m

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

Loss: 1.3108

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: 0.0006
train_batch_size: 64
eval_batch_size: 64
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 282
num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
2.0387	0.0708	1000	2.0164
1.7820	0.1417	2000	1.7784
1.6755	0.2125	3000	1.6811
1.5983	0.2833	4000	1.6234
1.5640	0.3542	5000	1.5827
1.5320	0.4250	6000	1.5524
1.5097	0.4958	7000	1.5257
1.4858	0.5667	8000	1.5055
1.4639	0.6375	9000	1.4870
1.4458	0.7083	10000	1.4707
1.4311	0.7791	11000	1.4541
1.4137	0.8500	12000	1.4401
1.3993	0.9208	13000	1.4267
1.3935	0.9916	14000	1.4146
1.3479	1.0625	15000	1.4014
1.3465	1.1333	16000	1.3898
1.3428	1.2041	17000	1.3794
1.3278	1.2750	18000	1.3677
1.3165	1.3458	19000	1.3582
1.3073	1.4166	20000	1.3480
1.2945	1.4875	21000	1.3394
1.2899	1.5583	22000	1.3310
1.2871	1.6291	23000	1.3247
1.2782	1.7000	24000	1.3196
1.2790	1.7708	25000	1.3152
1.2694	1.8416	26000	1.3126
1.2685	1.9125	27000	1.3111
1.2628	1.9833	28000	1.3108
1.2599	2.0	28236	1.3108

Framework versions

Transformers 5.19.0
Pytorch 2.11.0+cu130
Datasets 5.1.0
Tokenizers 0.23.3

Downloads last month: 546

Safetensors

Model size

97.7M params

Tensor type

F32

Model tree for laskar-ks/alcyone-base-97.7M

Finetunes

1 model