Upload README.md with huggingface_hub

README.md

@@ -104,22 +104,22 @@ printf 'Қазақша\nHello\n' | spellman detect --lines               # kaz /
 
 | eval | accuracy |
 |---|---|
-| held-out mix (239,042 rows, pristine test) | 98.30% |
-| Tatoeba (37,051, out-of-domain) | 98.42% |
+| held-out mix (238,985 rows, pristine test) | 98.31% |
+| Tatoeba (37,051, out-of-domain) | 98.39% |
 | Tatoeba granularity ladder (word / pair / triple) | 66.7 / 86.2 / 93.3 |
-| wild Russian tweets (2,606, label-audited; 61% @mentions, 20% URLs) | 92.56% |
-| COSMUS gold-labeled wild Russian (2,808, unseen) | 96.79% |
-| short-utterance referee (orthography-certified, ≤19 chars) | 89.90% |
+| wild Russian tweets (2,606, label-audited; 61% @mentions, 20% URLs) | 93.40% |
+| COSMUS gold-labeled wild Russian (2,808, unseen) | 96.72% |
+| short-utterance referee (orthography-certified, ≤19 chars) | 90.07% |
 | throughput | ~3.6 µs/sample (fp16 JIT plan, CPU) |
 
 On identical eval rows it beats fastText lid.176 (Tatoeba 94.90% vs
-98.42%) and lingua 1.8 high-accuracy on the shared languages at ~30x
+98.39%) and lingua 1.8 high-accuracy on the shared languages at ~30x
 its speed; whichlang and lingua cover only 10 and 17 of these 30
 classes. Against open-LID SOTA GlotLID v3 it wins the wild held-out
-workload (98.30 vs 93.58), the single-word rung (66.7 vs 43.9), the
-wild-Russian referee (92.56 vs 82.73) and the short-utterance referee
-(89.90 vs 71.25) while running ~100x faster; GlotLID's larger training
-set still leads clean out-of-domain sentences (99.25 vs 98.42).
+workload (98.31 vs 93.58), the single-word rung (66.6 vs 43.9), the
+wild-Russian referee (93.40 vs 82.73) and the short-utterance referee
+(90.07 vs 71.25) while running ~100x faster; GlotLID's larger training
+set still leads clean out-of-domain sentences (99.25 vs 98.39).
 
 ## Training data