Text Generation
Transformers
Safetensors
English
engram_qwen3
microllama
qwen3
engram
pretraining
small-language-model
custom_code
Instructions to use keeeeenw/MicroLlama2-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use keeeeenw/MicroLlama2-base with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="keeeeenw/MicroLlama2-base", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("keeeeenw/MicroLlama2-base", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use keeeeenw/MicroLlama2-base with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "keeeeenw/MicroLlama2-base" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "keeeeenw/MicroLlama2-base", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/keeeeenw/MicroLlama2-base
- SGLang
How to use keeeeenw/MicroLlama2-base 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 "keeeeenw/MicroLlama2-base" \ --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": "keeeeenw/MicroLlama2-base", "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 "keeeeenw/MicroLlama2-base" \ --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": "keeeeenw/MicroLlama2-base", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use keeeeenw/MicroLlama2-base with Docker Model Runner:
docker model run hf.co/keeeeenw/MicroLlama2-base
File size: 1,685 Bytes
1be6cef 9077e1f 1be6cef 9077e1f 1be6cef 9077e1f 1be6cef 9077e1f 1be6cef | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 | {
"schema_version": 2,
"release_date": "2026-10-30",
"checkpoint_label": "selected-c-6600m",
"step": 432299,
"valid_targets": 56602095602,
"added_valid_targets": 6602041425,
"checkpoint_manifest_sha256": "e0f00b28e1fd9ee11e63107b636f327d3e4fd2d11363b7d00cf66e8728098a25",
"weight_sha256": "f13afc6e011c745f4ac88ac055ca5d7551306218c3e361a51a26829e2cc1664b",
"weight_bytes": 1552148790,
"weight_changes": false,
"source_format": "native MLX production checkpoint; FP32 model tensors with integer Engram identity map",
"export_format": "Hugging Face inference-only safetensors; byte-identical model tensor file",
"training_context_length": 2048,
"inherited_max_position_embeddings": 8192,
"tokenizer_sha256": "df1d8d5ec5d091b460562ffd545e4a5e91d17d4a0db7ebe733be34ed374377bd",
"original_nvidia_parent": {
"source_repo": "keeeeenw/MicroLlama2-388M-checkpoints",
"source_revision": "527ab4101332284704765e1b2f3cfb577609bf56",
"source_prefix": "checkpoint/fineweb-edu-338m-8k-local-engram-4090/step-00381824",
"valid_targets": 50000054177
},
"previous_inference_revision": "e78d86500746b784ee6af0e91e6cc594b3366b8d",
"inference_metadata_changes": [
"Auto class remote-code mappings retained",
"Generation cache enabled; original inference generation policy retained",
"No optimizer, RNG, reader, training state, dataset or inherited chat template included"
],
"evaluation_results": "evaluation_results.json",
"evaluation_audits": {
"independent_full_audits": 2,
"both_passed": true,
"scorer_completed_normally": true
},
"training_hardware": "Apple M5 Ultra",
"pipeline": "Mac continuation pipeline"
}
|