Text Generation
Transformers
Safetensors
minspark
language-model
transformer
rope
gqa
custom_code
tiny
looped
slm
custom-architecture
custom-tokenizer
Instructions to use MinimaLabs/min-spark with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MinimaLabs/min-spark with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="MinimaLabs/min-spark", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("MinimaLabs/min-spark", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use MinimaLabs/min-spark with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "MinimaLabs/min-spark" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MinimaLabs/min-spark", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/MinimaLabs/min-spark
- SGLang
How to use MinimaLabs/min-spark 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 "MinimaLabs/min-spark" \ --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": "MinimaLabs/min-spark", "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 "MinimaLabs/min-spark" \ --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": "MinimaLabs/min-spark", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use MinimaLabs/min-spark with Docker Model Runner:
docker model run hf.co/MinimaLabs/min-spark
scrub internal project references from run_lmeval.py
Browse files- run_lmeval.py +3 -6
run_lmeval.py
CHANGED
|
@@ -1,8 +1,7 @@
|
|
| 1 |
-
"""run_lmeval.py — reproduce min-spark's published benchmarks
|
| 2 |
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
scoring rules. CLI: python run_lmeval.py --effort medium --tasks arc_easy,wikitext [--limit N]
|
| 6 |
"""
|
| 7 |
import argparse
|
| 8 |
import json
|
|
@@ -34,7 +33,6 @@ class MinSparkLM(TemplateLM):
|
|
| 34 |
self._device = device
|
| 35 |
self._max_length = max_length
|
| 36 |
|
| 37 |
-
# --- required TemplateLM/LM surface ---
|
| 38 |
@property
|
| 39 |
def eot_token_id(self) -> int:
|
| 40 |
return 2 # EOS
|
|
@@ -61,7 +59,6 @@ class MinSparkLM(TemplateLM):
|
|
| 61 |
def tok_decode(self, tokens, skip_special_tokens=True):
|
| 62 |
return self._tok.decode(tokens, skip_special_tokens=skip_special_tokens)
|
| 63 |
|
| 64 |
-
# --- scoring execution context (adapter-owned) ---
|
| 65 |
def _loglikelihood_tokens(self, requests, disable_tqdm=False, override_use_cache=None):
|
| 66 |
# Abstract in TemplateLM; unused (we override loglikelihood directly).
|
| 67 |
raise NotImplementedError("MinSparkLM overrides loglikelihood directly")
|
|
|
|
| 1 |
+
"""run_lmeval.py — reproduce min-spark's published benchmarks.
|
| 2 |
|
| 3 |
+
A LoglikelihoodLM (TemplateLM subclass) over MinSparkForCausalLM with the
|
| 4 |
+
repo's scoring rules. CLI: python run_lmeval.py --effort medium --tasks arc_easy,wikitext [--limit N]
|
|
|
|
| 5 |
"""
|
| 6 |
import argparse
|
| 7 |
import json
|
|
|
|
| 33 |
self._device = device
|
| 34 |
self._max_length = max_length
|
| 35 |
|
|
|
|
| 36 |
@property
|
| 37 |
def eot_token_id(self) -> int:
|
| 38 |
return 2 # EOS
|
|
|
|
| 59 |
def tok_decode(self, tokens, skip_special_tokens=True):
|
| 60 |
return self._tok.decode(tokens, skip_special_tokens=skip_special_tokens)
|
| 61 |
|
|
|
|
| 62 |
def _loglikelihood_tokens(self, requests, disable_tqdm=False, override_use_cache=None):
|
| 63 |
# Abstract in TemplateLM; unused (we override loglikelihood directly).
|
| 64 |
raise NotImplementedError("MinSparkLM overrides loglikelihood directly")
|