README: add v75 to arch tags + dirs

v75: replace bundle with freshly-converted build

Prune QHexRT HNPU bundle to runtime-minimum artifacts

Add HNPU arch tags to sanitized model card

Sanitize public model card metadata

Add chat template block (runtime-owned chat templating)

Add chat template block (runtime-owned chat templating)

Add chat template block (runtime-owned chat templating)

Upload lfm2_5_230m_v81.zip with huggingface_hub

Scrub internal engineering notes from manifest _comment

Scrub internal engineering notes from manifest _comment

Scrub internal engineering notes from manifest _comment

Add root config.json to enable Hub download counting

Model card: drop greedy-output line, keep it clean

Refresh hero: QHexRT × LFM 2.5 230M benchmark panel layout

Update hero thumbnail with LFM 2.5 230M model name

QHexRT launch: 100% NPU, multimodal roadmap, remove open-source refs

QHexRT launch article: full-stack NPU inference, not open source

Link full QHexRT benchmark write-up from model card

Add QHexRT benchmark article draft for Hub blog

Add QHexRT benchmark hero thumbnail

Add QHexRT benchmark chart: TTFT vs prompt

Add QHexRT benchmark chart: prefill throughput

Add v81 (SM8850) W8 bundle + CPU head-to-head to model card (#1)

Upload v79/tokenizer.json with huggingface_hub

Upload v79/lfm_embed_f16.bin with huggingface_hub

Upload v79/lfm230_lmh_w8.bin with huggingface_hub

Upload v79/lfm230_pf_2048_w8.bin with huggingface_hub

Upload v79/lfm230_pf_512_w8.bin with huggingface_hub

Upload v79/lfm230_dec_2048_w8.bin with huggingface_hub

Upload v79/lfm230_dec_512_w8.bin with huggingface_hub

Upload v79/lfm2-5-230m-2048.json with huggingface_hub

Upload v79/lfm2-5-230m.json with huggingface_hub

Upload README.md with huggingface_hub

initial commit