Activity
This seems to be working for dense Qwen3.5!!!
This seems to be working for dense Qwen3.5!!!
Autoparser: True streaming (#20177)
Autoparser: True streaming (#20177)
Autoparser - complete refactoring of parser architecture
Autoparser - complete refactoring of parser architecture
Force push
common : introduce composable PEG parser combinators for chat parsing…
common : introduce composable PEG parser combinators for chat parsing…
Force push
Autoparser - complete refactoring of parser architecture
Autoparser - complete refactoring of parser architecture
common : introduce composable PEG parser combinators for chat parsing…
common : introduce composable PEG parser combinators for chat parsing…
Force push
add back qwen_coder_xml and mirothinker
add back qwen_coder_xml and mirothinker
Force push
Minor: do not do SILU on the whole convolution output
Minor: do not do SILU on the whole convolution output
add back: common_chat_parse_qwen3_coder_xml
add back: common_chat_parse_qwen3_coder_xml
add back qwen_coder_xml and mirothinker
add back qwen_coder_xml and mirothinker
common : fix json schema with '\' in literals (#17307)
common : fix json schema with '\' in literals (#17307)
Fix split mode graph with Qwen3.5-MoE/Qwen3-Next hybryd inference
Fix split mode graph with Qwen3.5-MoE/Qwen3-Next hybryd inference
Disable split mode graph for recurrent/hybrid models when tensor over…
Disable split mode graph for recurrent/hybrid models when tensor over…
Pull request merge
Disable split mode graph for recurrent/hybrid models when tensor over…
Disable split mode graph for recurrent/hybrid models when tensor over…