Aider Polyglot results
#3
by
Fernanda24
- opened
Hi just wanted to share the Aider Polyglot benchmark results I got for this (with thinking disabled). It's near identical to the full GLM-4.6! Also question: Any chance we will see a Kimi-K2 Thinking REAP? :)
no_think
model: openai/glm-4.6-reap-268b-a32b-fp8
edit_format: diff
commit_hash: c74f5ef
pass_rate_1: 11.6
pass_rate_2: 39.1
pass_num_1: 26
pass_num_2: 88
percent_cases_well_formed: 92.4
error_outputs: 21
num_malformed_responses: 20
num_with_malformed_responses: 17
user_asks: 39
lazy_comments: 0
syntax_errors: 0
indentation_errors: 0```