ModelChorusModelChorus
ChangamotoMazungumzoOrodha ya NafasiTelezeshaTathminiHistoria
Jinsi inavyofanya kaziAPIMasharti ya HudumaSera ya Faragha

Hakimiliki 2026 MeetKai Inc.

Tathmini/GLM 5/Kazi zote

GLM 5

kazi 12

Kila safu hapa chini ni kazi moja ya tathmini ambayo muundo huu ulitathminiwa kwayo. Safuwima ya Alama hupima wastani wa vipimo vyote ambavyo kazi inaripoti (usahihi, F1, ulinganifu kamili, n.k.). Bofya safu ili kuvinjari maswali binafsi na majibu ya muundo.

Wastani
57.0
AlamaLughaKaziVipimo
87.3Swahili (Tanzania)
swahili_sib200
swahili classification
f1_macro: 87.3sample_len: 204.0
f1_macro: 87.3sample_len: 204.0
85.7English (US)
english_gsm8k
english math
exact_match: 85.7sample_len: 1319.0
exact_match: 85.7sample_len: 1319.0
84.9French (France)
french_sib200
french classification
f1_macro: 84.9sample_len: 204.0
f1_macro: 84.9sample_len: 204.0
84.0English (US)
english_mgsm
english math
exact_match: 84.0sample_len: 250.0
exact_match: 84.0sample_len: 250.0
80.8Swahili (Tanzania)
swahili_afrimgsm
swahili afrimgsm
exact_match: 80.8sample_len: 250.0
exact_match: 80.8sample_len: 250.0
73.5Spanish (Spain)
spanish_xquad_es
spanish xquad es
exact_match: 63.6f1: 83.4sample_len: 1190.0
exact_match: 63.6f1: 83.4sample_len: 1190.0
70.1Swahili (Tanzania)
swahili_afrixnli
swahili nli
f1_macro: 70.1sample_len: 600.0
f1_macro: 70.1sample_len: 600.0
50.6English (US)
english_mmlu_pro
english mmlu pro
exact_match: 50.6sample_len: 2100.0
exact_match: 50.6sample_len: 2100.0
36.3English (US)
ifeval
ifeval
inst_level_loose_acc: 45.0inst_level_strict_acc: 43.0prompt_level_loose_acc: 29.8prompt_level_strict_acc: 27.5sample_len: 541.0
inst_level_loose_acc: 45.0inst_level_strict_acc: 43.0prompt_level_loose_acc: 29.8prompt_level_strict_acc: 27.5sample_len: 541.0
11.5Spanish (Spain)
spanish_global_mmlu
spanish mcq
f1_macro: 11.5sample_len: 400.0
f1_macro: 11.5sample_len: 400.0
10.3Spanish (Spain)
spanish_belebele
spanish mcq
f1_macro: 10.3sample_len: 900.0
f1_macro: 10.3sample_len: 900.0
9.3English (US)
english_belebele
english mcq
f1_macro: 9.3sample_len: 900.0
f1_macro: 9.3sample_len: 900.0