docs: README fully in English (integrity notice + re-validation)

docs: integrity notice + re-validation (qkv scale-grid repair 2026-07-20)

fix: qkv_proj per-rank scale-grid repair (ranks 1..3 of all 48 layers)\n\nPre-fix converter dequantized fused qkv with a flat scale grid, corrupting\nranks 1..NR-1 (rank 0 intact). Repaired from XiaomiMiMo/MiMo-V2.5 fp8 source\nwith the fixed per-rank dequant + identical int8 requant. Verified:\nverify_mimo_qkv 5/5 layers PASS all ranks; ppl prose 9.97 / code 2.07\n(code matches the 12.40/2.07 reference axis); greedy 'The capital of France\nis' -> 'Paris.' See peng findings.md §46.

Add out-00015.safetensors (1.7 GB, int4/int8 peng container)

Add out-00014.safetensors (17.2 GB, int4/int8 peng container)

Add out-00013.safetensors (1.7 GB, int4/int8 peng container)

Add out-00012.safetensors (17.2 GB, int4/int8 peng container)

Add out-00011.safetensors (1.7 GB, int4/int8 peng container)

Add out-00010.safetensors (17.2 GB, int4/int8 peng container)

Add out-00009.safetensors (1.7 GB, int4/int8 peng container)

Add out-00008.safetensors (17.2 GB, int4/int8 peng container)

Add out-00007.safetensors (1.7 GB, int4/int8 peng container)

Add out-00006.safetensors (17.2 GB, int4/int8 peng container)

Validated: real-model canary passes; measured numbers; qkv de-interleave note

Add out-00005.safetensors (1.7 GB, int4/int8 peng container)

Add out-00004.safetensors (17.2 GB, int4/int8 peng container)

Add out-00003.safetensors (1.7 GB, int4/int8 peng container)

Add out-00002.safetensors (17.2 GB, int4/int8 peng container)

Add out-00001.safetensors (12.6 GB, int4/int8 peng container)

Add out-00000.safetensors (17.2 GB, int4/int8 peng container)

Add WIP model card

Add tokenizer_config.json

Add tokenizer.json

Add generation_config.json

Add config.json

initial commit