Instructions to use amazingvince/bitllama-goodwiki with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use amazingvince/bitllama-goodwiki with Transformers:

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

pipe = pipeline("text-generation", model="amazingvince/bitllama-goodwiki")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("amazingvince/bitllama-goodwiki")
model = AutoModelForCausalLM.from_pretrained("amazingvince/bitllama-goodwiki", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use amazingvince/bitllama-goodwiki with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/amazingvince/bitllama-goodwiki

SGLang

How to use amazingvince/bitllama-goodwiki 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 "amazingvince/bitllama-goodwiki" \
    --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": "amazingvince/bitllama-goodwiki",
		"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 "amazingvince/bitllama-goodwiki" \
        --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": "amazingvince/bitllama-goodwiki",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use amazingvince/bitllama-goodwiki with Docker Model Runner:
```
docker model run hf.co/amazingvince/bitllama-goodwiki
```

bitllama-goodwiki

This model was trained from scratch on the BEE-spoke-data/goodwiki-deduped-split dataset. It achieves the following results on the evaluation set:

Loss: 3.0525
Accuracy: 0.4285

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: 0.0008
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
6.1199	0.04	100	6.0749	0.1542
5.3869	0.07	200	5.3267	0.2032
4.9187	0.11	300	4.8566	0.2386
4.6185	0.14	400	4.5535	0.2624
4.3509	0.18	500	4.3388	0.2801
4.1666	0.21	600	4.1692	0.2956
4.0456	0.25	700	4.0399	0.3089
3.9273	0.28	800	3.9318	0.3193
3.8447	0.32	900	3.8173	0.3327
3.7143	0.35	1000	3.7108	0.3461
3.6485	0.39	1100	3.6116	0.3590
3.5171	0.42	1200	3.5303	0.3693
3.4464	0.46	1300	3.4554	0.3780
3.3955	0.49	1400	3.3999	0.3851
3.3551	0.53	1500	3.3432	0.3919
3.2787	0.56	1600	3.2981	0.3974
3.2705	0.6	1700	3.2566	0.4023
3.2281	0.64	1800	3.2172	0.4075
3.1759	0.67	1900	3.1826	0.4118
3.1603	0.71	2000	3.1547	0.4152
3.1328	0.74	2100	3.1283	0.4186
3.0916	0.78	2200	3.1055	0.4215
3.0939	0.81	2300	3.0875	0.4238
3.0584	0.85	2400	3.0732	0.4257
3.0711	0.88	2500	3.0631	0.4271
3.0612	0.92	2600	3.0565	0.4280
3.081	0.95	2700	3.0534	0.4284
3.0378	0.99	2800	3.0525	0.4285

Framework versions

Transformers 4.37.2
Pytorch 2.2.0+cu121
Datasets 2.16.1
Tokenizers 0.15.0

Downloads last month: 6

Safetensors

Model size

0.4B params

Tensor type

F32

Evaluation results

Accuracy on BEE-spoke-data/goodwiki-deduped-split
self-reported

0.429