Miril-Drone-2B-1-bnb8
8-bit CUDA deployment variant of Miril-Drone-2B-1
Drones can talk, and this is the lower-memory CUDA path.
This repository packages the 8-bit bitsandbytes variant of
Miril-Drone-2B-1, a 2B-class aerial VLM for drone-view imagery.
Use the primary model card for behavior, prompting, schemas, examples, WALDO vocabulary, limitations, and safety notes:
The V1 prompt contract is the same as the main model:
caption_v1,
simple_answer_v1, and
operational_coordinate_v2. V1 operational coordinates are rough representative grid cues for review, not flight-control commands.
Miril-DroneVLM-2B-2 is the newer plain-English typed-router generation.
Interactive demo:
Use
Use transformers>=5.12.1. This model retains Gemma 4 E2B's declared shared-KV layout; language layers 15 through 34 intentionally reuse key/value states and do not carry separate k_proj, v_proj, or k_norm tensors.
1python eval_generate.py \
2 --model_id MirilAI/Miril-Drone-2B-1-bnb8 \
3 --processor_id MirilAI/Miril-Drone-2B-1-bnb8 \
4 --jsonl eval.jsonl \
5 --image_root images \
6 --out_jsonl predictions.jsonl \
7 --batch_size 1 \
8 --max_new_tokens 256
Deployment Profile
MIRIL_VARIANT_PROFILE_PENDING
Complete Benchmark
MIRIL_RELEASE_METRICS_PENDING
The merged and quantized variants are evaluated on identical held-out cases. Automated scores are regression signals, not safety certification.
License
Apache License 2.0. See LICENSE and NOTICE.
