{ "text": "各位朋友大家好\n欢迎来关注昆蓬套\n那今天给大家介绍一个\nDeepseek V4 Flash 0731\n携带多摩泰的一个版本\n那这个是社区的大佬他们做的\n它是怎么实现的\n其实它是把Deepseek V4 Flash 0731\n这个模型在前面套了一层\n然后套的是什么呢\n我去扒了一下\n它应该是套的是\n千问3.535B的模型\n然后把它的多摩泰的那部分\n就是MUN VIT这个模块\n然后把它移植到\nDeepseek V4 Flash上来\n这样的话\nDeepseek V4 Flash\n就可以实现多摩泰的一个效果\n那其实前几期的视频\n我也给大家讲了一下\n怎么样把\n比如说通过Skill方式\n把那个节约的多摩泰\n或者是千问的多摩泰\n移植到Deepseek V4 Flash上来\n在Agent里面实现\n一个多摩泰的一个效果\n那那个是通过后期进行合成的\n那这个项目\n他们是通过\n大模型本身的能力上面\n进行合成的\n它相当于在\n上面套了一层视觉编码器\n当然它这个模型\n也支持Deepseek V4 Flash\n然后进行推理\n目前它的整个一个\n基准测试的一个速度\n还是非常成功的\n它这种合并\n就是达到了一个\n就是6.5的一个效果\n那说明了\n它的整个的一个模型的能力\n还是非常不错的\n那当然它这个的话\n对Deepseek V4 Flash本身的模型的\n损失是很小的\n几乎是可以忽略不尽\n就是它对模型本身\n没有做改造\n只是在它前面套了一层\n就是如果判断是\n图片的话就路由到\n它的视频编码器\n解码器上面去做一些工作\n那这个其实\n它的这个模型的能力\n虽然说不是很官方\n前段时间好像也给大家说过\n当然在我的群里面说过\n就是说JM5.2给它加上多摩泰的一个效果\n那Deepseek V4加上多摩泰\n它的那个使用的场景更多\n那这个模型呢\n目前呢\n我在网上看到有人在两台DGXBAR上跑\n有一个不错的一个效果\n就相当于它完全本地\n然后呢推你的成本极低的情况下\n还能实现一个多摩泰\n然后一照上下文的一个效果\n那这个模型还是非常非常不错的\n那我们其实呢\n社区呢有这样的说明\n这些还是有非常多的一个需求\n那目前呢\n它这个参数还是305B\n就是本身没有增加太多\n然后呢\n它目前是168G的一个大小\n所以说呢\n两台DGXBAR肯定是可以把它跑起来的\n当然在麦克上也能跑\n我目前并没有把它跑起来\n因为我对这样的一个\n就是本地去跑的需求不是很大\n因为如果有这样的话\n我已经通过我的那个Skills进行合并了\n那感兴趣的话呢\n大家可以来尝试一下\n那我觉得这种的话\n是一个非常有价值\n当然我们更希望\nDeepseek V4官方能够推出\n在多摩泰的一个效果\n那我们再期待一下吧\n那好的\n今天就介绍到这\n谢谢大家", "language": null, "duration": 160.124813, "segments": [ { "id": 0, "seek": 0, "start": 0.36, "end": 1.04, "text": "各位朋友大家好", "tokens": [ 50383, 22073, 19828, 15248, 50417 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 1, "seek": 0, "start": 1.04, "end": 1.86, "text": "欢迎来关注昆蓬套", "tokens": [ 50417, 28566, 17699, 6912, 28053, 26432, 1431, 228, 39198, 105, 1881, 245, 50458 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 2, "seek": 0, "start": 1.86, "end": 3.22, "text": "那今天给大家介绍一个", "tokens": [ 50458, 4184, 12074, 23197, 6868, 30312, 10115, 235, 20182, 50526 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 3, "seek": 0, "start": 3.22, "end": 4.76, "text": "Deepseek V4 Flash 0731", "tokens": [ 50526, 11089, 595, 405, 916, 691, 19, 20232, 1958, 22, 12967, 50603 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 4, "seek": 0, "start": 4.76, "end": 6.92, "text": "携带多摩泰的一个版本", "tokens": [ 50603, 22220, 118, 4845, 99, 6392, 34783, 102, 6847, 108, 1546, 20182, 42096, 8802, 50711 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 5, "seek": 0, "start": 6.92, "end": 9.14, "text": "那这个是社区的大佬他们做的", "tokens": [ 50711, 4184, 15368, 1541, 27658, 9937, 118, 1546, 3582, 1593, 105, 47911, 10907, 1546, 50822 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 6, "seek": 0, "start": 9.14, "end": 10.56, "text": "它是怎么实现的", "tokens": [ 50822, 11284, 1541, 15282, 24726, 20204, 1546, 50893 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 7, "seek": 0, "start": 10.56, "end": 13.32, "text": "其实它是把Deepseek V4 Flash 0731", "tokens": [ 50893, 41646, 11284, 1541, 16075, 11089, 595, 405, 916, 691, 19, 20232, 1958, 22, 12967, 51031 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 8, "seek": 0, "start": 13.32, "end": 15.08, "text": "这个模型在前面套了一层", "tokens": [ 51031, 15368, 41908, 39823, 3581, 8945, 8833, 1881, 245, 2289, 2257, 9636, 224, 51119 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 9, "seek": 0, "start": 15.08, "end": 16.32, "text": "然后套的是什么呢", "tokens": [ 51119, 26636, 1881, 245, 24620, 10440, 6240, 51181 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 10, "seek": 0, "start": 16.32, "end": 17.78, "text": "我去扒了一下", "tokens": [ 51181, 1654, 6734, 3416, 240, 2289, 8861, 51254 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 11, "seek": 0, "start": 17.78, "end": 20.04, "text": "它应该是套的是", "tokens": [ 51254, 11284, 44297, 44646, 1541, 1881, 245, 24620, 51367 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 12, "seek": 0, "start": 20.04, "end": 25.580000000000002, "text": "千问3.535B的模型", "tokens": [ 51367, 20787, 22064, 18, 13, 20, 8794, 33, 1546, 41908, 39823, 51644 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 13, "seek": 0, "start": 25.580000000000002, "end": 27.32, "text": "然后把它的多摩泰的那部分", "tokens": [ 51644, 26636, 42061, 1546, 6392, 34783, 102, 6847, 108, 1546, 4184, 32174, 51731 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 14, "seek": 0, "start": 27.32, "end": 29.5, "text": "就是MUN VIT这个模块", "tokens": [ 51731, 5620, 44, 3979, 691, 3927, 15368, 41908, 47734, 51840 ], "temperature": 0, "avg_logprob": -0.16290834090288947, "compression_ratio": 1.3137254901960784, "no_speech_prob": 2.5164578754122147e-11, "source": "whisper" }, { "id": 15, "seek": 2950, "start": 29.5, "end": 30.8, "text": "然后把它移植到", "tokens": [ 50365, 26636, 42061, 8204, 119, 38739, 235, 4511, 50430 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 16, "seek": 2950, "start": 30.8, "end": 32.78, "text": "Deepseek V4 Flash上来", "tokens": [ 50430, 11089, 595, 405, 916, 691, 19, 20232, 5708, 6912, 50529 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 17, "seek": 2950, "start": 32.78, "end": 33.26, "text": "这样的话", "tokens": [ 50529, 21209, 44575, 50553 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 18, "seek": 2950, "start": 33.26, "end": 34.12, "text": "Deepseek V4 Flash", "tokens": [ 50553, 11089, 595, 405, 916, 691, 19, 20232, 50596 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 19, "seek": 2950, "start": 34.12, "end": 35.96, "text": "就可以实现多摩泰的一个效果", "tokens": [ 50596, 36781, 24726, 20204, 6392, 34783, 102, 6847, 108, 1546, 20182, 43076, 9319, 50688 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 20, "seek": 2950, "start": 35.96, "end": 37.96, "text": "那其实前几期的视频", "tokens": [ 50688, 4184, 41646, 8945, 6336, 254, 16786, 1546, 40656, 39752, 50788 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 21, "seek": 2950, "start": 37.96, "end": 38.8, "text": "我也给大家讲了一下", "tokens": [ 50788, 27761, 23197, 6868, 39255, 2289, 8861, 50830 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 22, "seek": 2950, "start": 38.8, "end": 39.8, "text": "怎么样把", "tokens": [ 50830, 48200, 16075, 50880 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 23, "seek": 2950, "start": 39.8, "end": 41.94, "text": "比如说通过Skill方式", "tokens": [ 50880, 36757, 8090, 19550, 16866, 50, 34213, 9249, 27584, 50987 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 24, "seek": 2950, "start": 41.94, "end": 44.82, "text": "把那个节约的多摩泰", "tokens": [ 50987, 16075, 43083, 45161, 16853, 99, 1546, 6392, 34783, 102, 6847, 108, 51131 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 25, "seek": 2950, "start": 44.82, "end": 46.3, "text": "或者是千问的多摩泰", "tokens": [ 51131, 31148, 1541, 20787, 22064, 1546, 6392, 34783, 102, 6847, 108, 51205 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 26, "seek": 2950, "start": 46.3, "end": 47.980000000000004, "text": "移植到Deepseek V4 Flash上来", "tokens": [ 51205, 8204, 119, 38739, 235, 4511, 11089, 595, 405, 916, 691, 19, 20232, 5708, 6912, 51289 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 27, "seek": 2950, "start": 47.980000000000004, "end": 49.08, "text": "在Agent里面实现", "tokens": [ 51289, 3581, 32, 6930, 15759, 8833, 24726, 20204, 51344 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 28, "seek": 2950, "start": 49.08, "end": 50.42, "text": "一个多摩泰的一个效果", "tokens": [ 51344, 20182, 6392, 34783, 102, 6847, 108, 1546, 20182, 43076, 9319, 51411 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 29, "seek": 2950, "start": 50.42, "end": 52.28, "text": "那那个是通过后期进行合成的", "tokens": [ 51411, 4184, 43083, 1541, 19550, 16866, 13547, 16786, 36700, 8082, 14245, 11336, 1546, 51504 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 30, "seek": 2950, "start": 52.28, "end": 52.980000000000004, "text": "那这个项目", "tokens": [ 51504, 4184, 15368, 10178, 117, 11386, 51539 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 31, "seek": 2950, "start": 52.980000000000004, "end": 53.900000000000006, "text": "他们是通过", "tokens": [ 51539, 47911, 1541, 19550, 16866, 51585 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 32, "seek": 2950, "start": 53.900000000000006, "end": 56.34, "text": "大模型本身的能力上面", "tokens": [ 51585, 3582, 41908, 39823, 8802, 19847, 1546, 8225, 13486, 49750, 51707 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 33, "seek": 2950, "start": 56.34, "end": 56.96, "text": "进行合成的", "tokens": [ 51707, 36700, 8082, 14245, 11336, 1546, 51738 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 34, "seek": 2950, "start": 56.96, "end": 57.5, "text": "它相当于在", "tokens": [ 51738, 11284, 15106, 16233, 37732, 3581, 51765 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 35, "seek": 2950, "start": 57.5, "end": 59.24, "text": "上面套了一层视觉编码器", "tokens": [ 51765, 49750, 1881, 245, 2289, 2257, 9636, 224, 40656, 24447, 38109, 244, 23230, 223, 34386, 51852 ], "temperature": 0, "avg_logprob": -0.11508519694490253, "compression_ratio": 1.485207100591716, "no_speech_prob": 9.979600479326223e-12, "source": "whisper" }, { "id": 36, "seek": 5924, "start": 59.24, "end": 61.06, "text": "当然它这个模型", "tokens": [ 50365, 40486, 11284, 15368, 41908, 39823, 50456 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 37, "seek": 5924, "start": 61.06, "end": 62.58, "text": "也支持Deepseek V4 Flash", "tokens": [ 50456, 6404, 35488, 11089, 595, 405, 916, 691, 19, 20232, 50532 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 38, "seek": 5924, "start": 62.58, "end": 63.58, "text": "然后进行推理", "tokens": [ 50532, 26636, 36700, 8082, 33597, 13876, 50582 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 39, "seek": 5924, "start": 63.58, "end": 65.36, "text": "目前它的整个一个", "tokens": [ 50582, 39004, 45224, 27662, 7549, 20182, 50671 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 40, "seek": 5924, "start": 65.36, "end": 66.58, "text": "基准测试的一个速度", "tokens": [ 50671, 26008, 6336, 228, 11038, 233, 5233, 243, 1546, 20182, 31217, 13127, 50732 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 41, "seek": 5924, "start": 66.58, "end": 69.6, "text": "还是非常成功的", "tokens": [ 50732, 45726, 14392, 11336, 32311, 1546, 50883 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 42, "seek": 5924, "start": 69.6, "end": 70.16, "text": "它这种合并", "tokens": [ 50883, 11284, 5562, 39810, 14245, 3509, 114, 50911 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 43, "seek": 5924, "start": 70.16, "end": 71.14, "text": "就是达到了一个", "tokens": [ 50911, 5620, 9830, 122, 21381, 20182, 50960 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 44, "seek": 5924, "start": 71.14, "end": 72.76, "text": "就是6.5的一个效果", "tokens": [ 50960, 5620, 21, 13, 20, 1546, 20182, 43076, 9319, 51041 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 45, "seek": 5924, "start": 72.76, "end": 73.64, "text": "那说明了", "tokens": [ 51041, 4184, 8090, 11100, 2289, 51085 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 46, "seek": 5924, "start": 73.64, "end": 75.8, "text": "它的整个的一个模型的能力", "tokens": [ 51085, 45224, 27662, 7549, 1546, 20182, 41908, 39823, 1546, 8225, 13486, 51193 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 47, "seek": 5924, "start": 75.8, "end": 77.24000000000001, "text": "还是非常不错的", "tokens": [ 51193, 45726, 14392, 1960, 29900, 1546, 51265 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 48, "seek": 5924, "start": 77.24000000000001, "end": 78.16, "text": "那当然它这个的话", "tokens": [ 51265, 4184, 40486, 11284, 15368, 44575, 51311 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 49, "seek": 5924, "start": 78.16, "end": 80.18, "text": "对Deepseek V4 Flash本身的模型的", "tokens": [ 51311, 8713, 11089, 595, 405, 916, 691, 19, 20232, 8802, 19847, 1546, 41908, 39823, 1546, 51412 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 50, "seek": 5924, "start": 80.18, "end": 81.38, "text": "损失是很小的", "tokens": [ 51412, 26075, 253, 28564, 1541, 4563, 7322, 1546, 51472 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 51, "seek": 5924, "start": 81.38, "end": 82.3, "text": "几乎是可以忽略不尽", "tokens": [ 51472, 6336, 254, 2930, 236, 1541, 6723, 4263, 121, 6904, 98, 1960, 1530, 121, 51518 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 52, "seek": 5924, "start": 82.3, "end": 83.5, "text": "就是它对模型本身", "tokens": [ 51518, 5620, 11284, 8713, 41908, 39823, 8802, 19847, 51578 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 53, "seek": 5924, "start": 83.5, "end": 84.42, "text": "没有做改造", "tokens": [ 51578, 17944, 10907, 34490, 37583, 51624 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 54, "seek": 5924, "start": 84.42, "end": 85.82000000000001, "text": "只是在它前面套了一层", "tokens": [ 51624, 36859, 3581, 11284, 8945, 8833, 1881, 245, 2289, 2257, 9636, 224, 51694 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 55, "seek": 5924, "start": 85.82000000000001, "end": 86.68, "text": "就是如果判断是", "tokens": [ 51694, 5620, 13119, 2437, 97, 4307, 255, 1541, 51737 ], "temperature": 0, "avg_logprob": -0.13872173999218232, "compression_ratio": 1.442367601246106, "no_speech_prob": 1.877856203869044e-11, "source": "whisper" }, { "id": 56, "seek": 8668, "start": 86.68, "end": 88.56, "text": "图片的话就路由到", "tokens": [ 50365, 3919, 122, 16668, 44575, 3111, 24658, 23786, 4511, 50459 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 57, "seek": 8668, "start": 88.56, "end": 90.64, "text": "它的视频编码器", "tokens": [ 50459, 45224, 40656, 39752, 38109, 244, 23230, 223, 34386, 50563 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 58, "seek": 8668, "start": 90.64, "end": 92.76, "text": "解码器上面去做一些工作", "tokens": [ 50563, 17278, 23230, 223, 34386, 49750, 6734, 10907, 38515, 41315, 50669 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 59, "seek": 8668, "start": 92.76, "end": 94.7, "text": "那这个其实", "tokens": [ 50669, 4184, 15368, 41646, 50766 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 60, "seek": 8668, "start": 94.7, "end": 96.4, "text": "它的这个模型的能力", "tokens": [ 50766, 45224, 15368, 41908, 39823, 1546, 8225, 13486, 50851 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 61, "seek": 8668, "start": 96.4, "end": 98.74000000000001, "text": "虽然说不是很官方", "tokens": [ 50851, 12026, 121, 5823, 8090, 7296, 4563, 31929, 9249, 50968 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 62, "seek": 8668, "start": 98.74000000000001, "end": 100.96000000000001, "text": "前段时间好像也给大家说过", "tokens": [ 50968, 8945, 28427, 44848, 33242, 6404, 23197, 6868, 8090, 16866, 51079 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 63, "seek": 8668, "start": 100.96000000000001, "end": 102.52000000000001, "text": "当然在我的群里面说过", "tokens": [ 51079, 40486, 3581, 14200, 7248, 97, 15759, 8833, 8090, 16866, 51157 ], "temperature": 0, "avg_logprob": -0.13118998612029642, "compression_ratio": 1.2138728323699421, "no_speech_prob": 1.7993292619755685e-11, "source": "whisper" }, { "id": 64, "seek": 10252, "start": 102.52, "end": 105.46, "text": "就是说JM5.2给它加上多摩泰的一个效果", "tokens": [ 50365, 5620, 8090, 41, 44, 20, 13, 17, 23197, 11284, 9990, 5708, 6392, 34783, 102, 6847, 108, 1546, 20182, 43076, 9319, 50512 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 65, "seek": 10252, "start": 105.46, "end": 106.92, "text": "那Deepseek V4加上多摩泰", "tokens": [ 50512, 4184, 11089, 595, 405, 916, 691, 19, 9990, 5708, 6392, 34783, 102, 6847, 108, 50585 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 66, "seek": 10252, "start": 106.92, "end": 108.75999999999999, "text": "它的那个使用的场景更多", "tokens": [ 50585, 45224, 43083, 22982, 9254, 1546, 50255, 50218, 19002, 6392, 50677 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 67, "seek": 10252, "start": 108.75999999999999, "end": 109.67999999999999, "text": "那这个模型呢", "tokens": [ 50677, 4184, 15368, 41908, 39823, 6240, 50723 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 68, "seek": 10252, "start": 109.67999999999999, "end": 110.08, "text": "目前呢", "tokens": [ 50723, 39004, 6240, 50743 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 69, "seek": 10252, "start": 110.08, "end": 112.44, "text": "我在网上看到有人在两台DGXBAR上跑", "tokens": [ 50743, 40528, 16469, 239, 5708, 18032, 45820, 3581, 36257, 15433, 35, 38, 55, 33, 1899, 5708, 32585, 50861 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 70, "seek": 10252, "start": 112.44, "end": 113.88, "text": "有一个不错的一个效果", "tokens": [ 50861, 2412, 20182, 1960, 29900, 1546, 20182, 43076, 9319, 50933 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 71, "seek": 10252, "start": 113.88, "end": 115.03999999999999, "text": "就相当于它完全本地", "tokens": [ 50933, 3111, 15106, 16233, 37732, 11284, 37100, 8802, 10928, 50991 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 72, "seek": 10252, "start": 115.03999999999999, "end": 117.34, "text": "然后呢推你的成本极低的情况下", "tokens": [ 50991, 26636, 6240, 33597, 18961, 11336, 8802, 7360, 223, 41377, 1546, 46514, 4438, 51106 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 73, "seek": 10252, "start": 117.34, "end": 118.64, "text": "还能实现一个多摩泰", "tokens": [ 51106, 14852, 8225, 24726, 20204, 20182, 6392, 34783, 102, 6847, 108, 51171 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 74, "seek": 10252, "start": 118.64, "end": 119.96, "text": "然后一照上下文的一个效果", "tokens": [ 51171, 26636, 2257, 32150, 5708, 4438, 17174, 1546, 20182, 43076, 9319, 51237 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 75, "seek": 10252, "start": 119.96, "end": 121.75999999999999, "text": "那这个模型还是非常非常不错的", "tokens": [ 51237, 4184, 15368, 41908, 39823, 45726, 14392, 14392, 1960, 29900, 1546, 51327 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 76, "seek": 10252, "start": 121.75999999999999, "end": 123.39999999999999, "text": "那我们其实呢", "tokens": [ 51327, 4184, 15003, 41646, 6240, 51409 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 77, "seek": 10252, "start": 123.39999999999999, "end": 126.67999999999999, "text": "社区呢有这样的说明", "tokens": [ 51409, 27658, 9937, 118, 6240, 2412, 21209, 1546, 8090, 11100, 51573 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 78, "seek": 10252, "start": 126.67999999999999, "end": 129.16, "text": "这些还是有非常多的一个需求", "tokens": [ 51573, 5562, 13824, 45726, 2412, 14392, 6392, 1546, 20182, 32535, 32718, 51697 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 79, "seek": 10252, "start": 129.16, "end": 130.06, "text": "那目前呢", "tokens": [ 51697, 4184, 39004, 6240, 51742 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 80, "seek": 10252, "start": 130.06, "end": 131.34, "text": "它这个参数还是305B", "tokens": [ 51742, 11284, 15368, 2129, 224, 33188, 45726, 3446, 20, 33, 51806 ], "temperature": 0, "avg_logprob": -0.1564959208170573, "compression_ratio": 1.4668587896253602, "no_speech_prob": 1.6469669378293972e-11, "source": "whisper" }, { "id": 81, "seek": 13134, "start": 131.34, "end": 133.52, "text": "就是本身没有增加太多", "tokens": [ 50365, 5620, 8802, 19847, 17944, 24228, 252, 9990, 9455, 6392, 50474 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 82, "seek": 13134, "start": 133.52, "end": 133.92000000000002, "text": "然后呢", "tokens": [ 50474, 26636, 6240, 50494 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 83, "seek": 13134, "start": 133.92000000000002, "end": 136.42000000000002, "text": "它目前是168G的一个大小", "tokens": [ 50494, 11284, 39004, 1541, 6866, 23, 38, 1546, 20182, 3582, 7322, 50619 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 84, "seek": 13134, "start": 136.42000000000002, "end": 137.34, "text": "所以说呢", "tokens": [ 50619, 7239, 8090, 6240, 50665 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 85, "seek": 13134, "start": 137.34, "end": 139.38, "text": "两台DGXBAR肯定是可以把它跑起来的", "tokens": [ 50665, 36257, 15433, 35, 38, 55, 33, 1899, 14356, 107, 12088, 1541, 6723, 42061, 32585, 9147, 6912, 1546, 50767 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 86, "seek": 13134, "start": 139.38, "end": 140.18, "text": "当然在麦克上也能跑", "tokens": [ 50767, 40486, 3581, 3994, 99, 24881, 5708, 6404, 8225, 32585, 50807 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 87, "seek": 13134, "start": 140.18, "end": 141.56, "text": "我目前并没有把它跑起来", "tokens": [ 50807, 1654, 39004, 3509, 114, 17944, 42061, 32585, 9147, 6912, 50876 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 88, "seek": 13134, "start": 141.56, "end": 142.92000000000002, "text": "因为我对这样的一个", "tokens": [ 50876, 34627, 1654, 8713, 21209, 1546, 20182, 50944 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 89, "seek": 13134, "start": 142.92000000000002, "end": 144.78, "text": "就是本地去跑的需求不是很大", "tokens": [ 50944, 5620, 8802, 10928, 6734, 32585, 1546, 32535, 32718, 7296, 4563, 3582, 51037 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 90, "seek": 13134, "start": 144.78, "end": 146.22, "text": "因为如果有这样的话", "tokens": [ 51037, 34627, 13119, 2412, 21209, 44575, 51109 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 91, "seek": 13134, "start": 146.22, "end": 147.98000000000002, "text": "我已经通过我的那个Skills进行合并了", "tokens": [ 51109, 1654, 49161, 19550, 16866, 14200, 43083, 50, 74, 2565, 36700, 8082, 14245, 3509, 114, 2289, 51197 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 92, "seek": 13134, "start": 147.98000000000002, "end": 148.72, "text": "那感兴趣的话呢", "tokens": [ 51197, 4184, 9709, 2347, 112, 39835, 44575, 6240, 51234 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 93, "seek": 13134, "start": 148.72, "end": 149.48000000000002, "text": "大家可以来尝试一下", "tokens": [ 51234, 6868, 6723, 6912, 1530, 251, 5233, 243, 8861, 51272 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 94, "seek": 13134, "start": 149.48000000000002, "end": 150.76, "text": "那我觉得这种的话", "tokens": [ 51272, 4184, 45681, 5562, 39810, 44575, 51336 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 95, "seek": 13134, "start": 150.76, "end": 151.68, "text": "是一个非常有价值", "tokens": [ 51336, 1541, 20182, 14392, 2412, 1550, 115, 40242, 51382 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 96, "seek": 13134, "start": 151.68, "end": 152.52, "text": "当然我们更希望", "tokens": [ 51382, 40486, 15003, 19002, 29955, 51424 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 97, "seek": 13134, "start": 152.52, "end": 154.34, "text": "Deepseek V4官方能够推出", "tokens": [ 51424, 11089, 595, 405, 916, 691, 19, 31929, 9249, 8225, 1787, 253, 33597, 7781, 51515 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 98, "seek": 13134, "start": 154.34, "end": 155.56, "text": "在多摩泰的一个效果", "tokens": [ 51515, 3581, 6392, 34783, 102, 6847, 108, 1546, 20182, 43076, 9319, 51576 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 99, "seek": 13134, "start": 155.56, "end": 157.16, "text": "那我们再期待一下吧", "tokens": [ 51576, 4184, 15003, 8623, 16786, 18390, 8861, 6062, 51656 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 100, "seek": 13134, "start": 157.16, "end": 157.64000000000001, "text": "那好的", "tokens": [ 51656, 4184, 20715, 51680 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 101, "seek": 13134, "start": 157.64000000000001, "end": 158.64000000000001, "text": "今天就介绍到这", "tokens": [ 51680, 12074, 3111, 30312, 10115, 235, 4511, 5562, 51730 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" }, { "id": 102, "seek": 13134, "start": 158.64000000000001, "end": 159.0, "text": "谢谢大家", "tokens": [ 51730, 19879, 6868, 51748 ], "temperature": 0, "avg_logprob": -0.10744232106431623, "compression_ratio": 1.3970588235294117, "no_speech_prob": 1.3116478389529895e-11, "source": "whisper" } ], "transcript_selection": { "schema": "videonote.transcript_selection.v1", "status": "SELECTED", "requested_language": null, "duration": 160.124813, "reference_path": null, "reference_kind": null, "reference_language": null, "selected_source": "whisper_segmented", "decision_reason": "source_unavailable_or_untrusted_whisper_passed", "whisper_rejected": false, "source_profile": { "segment_count": 0, "duration_coverage": 0.0, "usable_full": false, "usable_partial": false, "hallucination_region_count": 0, "hallucination_regions": [], "language": { "requested": null, "matched": false } }, "whisper_profile": { "segment_count": 103, "multilingual": false, "first_start": 0.36, "final_end": 159.0, "duration_coverage": 0.992975, "maximum_gap_seconds": 0.36, "inverted_count": 0, "language": { "requested": null, "matched": true, "latin": 155, "han": 816, "kana": 0, "hangul": 0 }, "hallucination_region_count": 0, "hallucination_regions": [], "usable_full": true, "usable_partial": true }, "comparison": { "window_seconds": 30.0, "compared_window_count": 0, "median_score": 0.0, "divergent_window_count": 0, "divergent_ratio": 0.0, "divergent_windows": [] }, "timestamp_alignment": { "applied": false, "estimated_offset_seconds": 0.0, "matched_segment_count": 0, "match_scores": [] }, "filled_source_profile": null, "whisper_duration_guard": { "media_duration": 160.125, "rejected_segment_count": 1, "rejected_segments": [ { "index": 103, "start": 159.0, "end": 188.98, "reason": "material_end_timestamp_overrun" } ] } } }