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
Download checkpoint_source.json from keeeeenw/MicroLlama2-base: direct link, hf CLI and curl.
- Browser
- Download file 1.69 kB
-
https://huggingface.co/keeeeenw/MicroLlama2-base/resolve/main/checkpoint_source.json
- Command line
-
hf download hf://keeeeenw/MicroLlama2-base/checkpoint_source.json
-
curl -L -o checkpoint_source.json https://huggingface.co/keeeeenw/MicroLlama2-base/resolve/main/checkpoint_source.json
1.69 kB
| { | |
| "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" | |
| } | |