legolas-zeng
legolas-zeng
比如说服务端挂了,客户端会一直等到客户端上线,然后重连。
# github.com/marten-seemann/qtls src\github.com\marten-seemann\qtls\common.go:1380:20: undefined: cpu.HasAES src\github.com\marten-seemann\qtls\common.go:1380:34: undefined: cpu.ARM64 src\github.com\marten-seemann\qtls\key_schedule.go:40:9: undefined: hkdf.Extract src\github.com\marten-seemann\qtls\key_schedule.go:59:12: undefined: hkdf.Expand src\github.com\marten-seemann\qtls\key_schedule.go:129:30: undefined: curve25519.ScalarSize src\github.com\marten-seemann\qtls\key_schedule.go:133:21: undefined: curve25519.X25519 src\github.com\marten-seemann\qtls\key_schedule.go:133:51: undefined: curve25519.Basepoint src\github.com\marten-seemann\qtls\key_schedule.go:212:20: undefined: curve25519.X25519
经过我多天的测试,web did跟mobile did很容易混淆写入代码里面,建议分开配置。
 as shown in figure,I want to show Chinese, but it will cause confusion
比如模型 Qwen3-235B-A22B ,支持think和非think模式,使用vllm,可以加参数 enable_thinking=True 来控制是否使用think,那么eino呢?要怎么配置?