Instructions to use SupraLabs/Supra-Mini-0.1M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SupraLabs/Supra-Mini-0.1M with Transformers:

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

pipe = pipeline("text-generation", model="SupraLabs/Supra-Mini-0.1M")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SupraLabs/Supra-Mini-0.1M")
model = AutoModelForCausalLM.from_pretrained("SupraLabs/Supra-Mini-0.1M")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use SupraLabs/Supra-Mini-0.1M with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SupraLabs/Supra-Mini-0.1M"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SupraLabs/Supra-Mini-0.1M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/SupraLabs/Supra-Mini-0.1M

SGLang

How to use SupraLabs/Supra-Mini-0.1M 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 "SupraLabs/Supra-Mini-0.1M" \
    --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": "SupraLabs/Supra-Mini-0.1M",
		"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 "SupraLabs/Supra-Mini-0.1M" \
        --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": "SupraLabs/Supra-Mini-0.1M",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use SupraLabs/Supra-Mini-0.1M with Docker Model Runner:
```
docker model run hf.co/SupraLabs/Supra-Mini-0.1M
```

Supra-Mini-0.1M / benchmarks.md

LH-Tech-AI

Rename benchmarks.sh to benchmarks.md

2c43df0 verified 12 days ago

preview code

raw

history blame contribute delete

10.2 kB

Tasks	Version	Filter	Metric		Value		Stderr
blimp	2	none	acc	↑	0.5177	±	0.0017
- blimp_adjunct_island	1	none	acc	↑	0.7430	±	0.0138
- blimp_anaphor_gender_agreement	1	none	acc	↑	0.2600	±	0.0139
- blimp_anaphor_number_agreement	1	none	acc	↑	0.4650	±	0.0158
- blimp_animate_subject_passive	1	none	acc	↑	0.5740	±	0.0156
- blimp_animate_subject_trans	1	none	acc	↑	0.6820	±	0.0147
- blimp_causative	1	none	acc	↑	0.4270	±	0.0156
- blimp_complex_NP_island	1	none	acc	↑	0.4380	±	0.0157
- blimp_coordinate_structure_constraint_complex_left_branch	1	none	acc	↑	0.0860	±	0.0089
- blimp_coordinate_structure_constraint_object_extraction	1	none	acc	↑	0.5060	±	0.0158
- blimp_determiner_noun_agreement_1	1	none	acc	↑	0.5960	±	0.0155
- blimp_determiner_noun_agreement_2	1	none	acc	↑	0.5470	±	0.0157
- blimp_determiner_noun_agreement_irregular_1	1	none	acc	↑	0.5110	±	0.0158
- blimp_determiner_noun_agreement_irregular_2	1	none	acc	↑	0.5840	±	0.0156
- blimp_determiner_noun_agreement_with_adj_2	1	none	acc	↑	0.4880	±	0.0158
- blimp_determiner_noun_agreement_with_adj_irregular_1	1	none	acc	↑	0.4500	±	0.0157
- blimp_determiner_noun_agreement_with_adj_irregular_2	1	none	acc	↑	0.5310	±	0.0158
- blimp_determiner_noun_agreement_with_adjective_1	1	none	acc	↑	0.5190	±	0.0158
- blimp_distractor_agreement_relational_noun	1	none	acc	↑	0.3480	±	0.0151
- blimp_distractor_agreement_relative_clause	1	none	acc	↑	0.3440	±	0.0150
- blimp_drop_argument	1	none	acc	↑	0.7320	±	0.0140
- blimp_ellipsis_n_bar_1	1	none	acc	↑	0.2240	±	0.0132
- blimp_ellipsis_n_bar_2	1	none	acc	↑	0.2920	±	0.0144
- blimp_existential_there_object_raising	1	none	acc	↑	0.7300	±	0.0140
- blimp_existential_there_quantifiers_1	1	none	acc	↑	0.7110	±	0.0143
- blimp_existential_there_quantifiers_2	1	none	acc	↑	0.0400	±	0.0062
- blimp_existential_there_subject_raising	1	none	acc	↑	0.6460	±	0.0151
- blimp_expletive_it_object_raising	1	none	acc	↑	0.6440	±	0.0151
- blimp_inchoative	1	none	acc	↑	0.3790	±	0.0153
- blimp_intransitive	1	none	acc	↑	0.5630	±	0.0157
- blimp_irregular_past_participle_adjectives	1	none	acc	↑	0.4000	±	0.0155
- blimp_irregular_past_participle_verbs	1	none	acc	↑	0.5430	±	0.0158
- blimp_irregular_plural_subject_verb_agreement_1	1	none	acc	↑	0.4460	±	0.0157
- blimp_irregular_plural_subject_verb_agreement_2	1	none	acc	↑	0.5100	±	0.0158
- blimp_left_branch_island_echo_question	1	none	acc	↑	0.8390	±	0.0116
- blimp_left_branch_island_simple_question	1	none	acc	↑	0.1170	±	0.0102
- blimp_matrix_question_npi_licensor_present	1	none	acc	↑	0.0020	±	0.0014
- blimp_npi_present_1	1	none	acc	↑	0.5060	±	0.0158
- blimp_npi_present_2	1	none	acc	↑	0.5070	±	0.0158
- blimp_only_npi_licensor_present	1	none	acc	↑	0.1620	±	0.0117
- blimp_only_npi_scope	1	none	acc	↑	0.0930	±	0.0092
- blimp_passive_1	1	none	acc	↑	0.5950	±	0.0155
- blimp_passive_2	1	none	acc	↑	0.6130	±	0.0154
- blimp_principle_A_c_command	1	none	acc	↑	0.5840	±	0.0156
- blimp_principle_A_case_1	1	none	acc	↑	0.9990	±	0.0010
- blimp_principle_A_case_2	1	none	acc	↑	0.4280	±	0.0157
- blimp_principle_A_domain_1	1	none	acc	↑	1.0000	±	0
- blimp_principle_A_domain_2	1	none	acc	↑	0.6010	±	0.0155
- blimp_principle_A_domain_3	1	none	acc	↑	0.5150	±	0.0158
- blimp_principle_A_reconstruction	1	none	acc	↑	0.1900	±	0.0124
- blimp_regular_plural_subject_verb_agreement_1	1	none	acc	↑	0.6880	±	0.0147
- blimp_regular_plural_subject_verb_agreement_2	1	none	acc	↑	0.5920	±	0.0155
- blimp_sentential_negation_npi_licensor_present	1	none	acc	↑	0.9990	±	0.0010
- blimp_sentential_negation_npi_scope	1	none	acc	↑	0.5420	±	0.0158
- blimp_sentential_subject_island	1	none	acc	↑	0.3570	±	0.0152
- blimp_superlative_quantifiers_1	1	none	acc	↑	0.4970	±	0.0158
- blimp_superlative_quantifiers_2	1	none	acc	↑	0.6980	±	0.0145
- blimp_tough_vs_raising_1	1	none	acc	↑	0.2810	±	0.0142
- blimp_tough_vs_raising_2	1	none	acc	↑	0.7660	±	0.0134
- blimp_transitive	1	none	acc	↑	0.6110	±	0.0154
- blimp_wh_island	1	none	acc	↑	0.2680	±	0.0140
- blimp_wh_questions_object_gap	1	none	acc	↑	0.7850	±	0.0130
- blimp_wh_questions_subject_gap	1	none	acc	↑	0.9600	±	0.0062
- blimp_wh_questions_subject_gap_long_distance	1	none	acc	↑	0.9490	±	0.0070
- blimp_wh_vs_that_no_gap	1	none	acc	↑	0.9830	±	0.0041
- blimp_wh_vs_that_no_gap_long_distance	1	none	acc	↑	0.9770	±	0.0047
- blimp_wh_vs_that_with_gap	1	none	acc	↑	0.0070	±	0.0026
- blimp_wh_vs_that_with_gap_long_distance	1	none	acc	↑	0.0190	±	0.0043
arc_easy	1	none	acc	↑	0.2639	±	0.0090
		none	acc_norm	↑	0.2731	±	0.0091
wikitext	2	none	bits_per_byte	↓	4.6536	±	N/A
		none	byte_perplexity	↓	25.1691	±	N/A
		none	word_perplexity	↓	30979484.4095	±	N/A

Groups	Version	Filter	n-shot	Metric		Value		Stderr
blimp	2	none	0	acc	↑	0.5177	±	0.0017