{"text": "各位朋友大家好,歡迎大家關注昆蓬特,今天給大家介紹一個小模型叫SoloX DuPRO 0.6B,那這個模型呢它是一個全商工的語音交互,什麼叫全商工就是能夠實時對話的一個模型,它的模型只有0.6B,0.6B什麼概念?就是你在手機上就能運行一個全本地的一個對話的一個模型,那它的效果還是非常不錯的,一會給大家演示一下,我本地沒有去跑它。首先第一個呢,它是用來還是VD模型的這種,然後它後面呢是用一個叫它那個語音模型是用了千問3的一個0.6B的一個模型應該是。然後它安裝我們可以看一下它的安裝的方式,它目前是Github一個倉庫,然後在Github上可以去運行推理它,因為目前對於這種全商工的這種推理的框架不多,所以比如說NAMA.CPP啊,它可能也還沒有支持這種模型。那前段時間,OpenBnB他們開玩那個叫NAMA.CPP-OMINI的一個模型,那它那個裡面是能實現這些功能,但是那個框架的可能也是OpenBnB他們剛出的,然後不太通用。那這個模型呢,它的信息很少,那我們可以看一下,它底下呢,說它致謝了一下這個千問的JM,千問的這個LM和JM4的,這應該是個TTS吧。這相當於它是一個縫合怪,那本身呢,基於這種全商工呢,它其實就是,第一個呢,是做語音識別,類似於SR,然後呢,然後識別完了之後呢,底層還有一個LM模型,模型進行推理。推理完成之後呢,再用TTS給你讀出來,就是這麼一個情況,然後只不過它處理速度極快,導致你以為是跟一個,呃,完整的一個模型進行對話。那這個是它的一個Demo,我們來演示一下。你好,你是誰?嗯,占用時間了,咱們繼續。嗯,抱歉,可能,我的麥克風被這個占用了,那我。我聽著呢。哦,不是,還是喊過來,我以為是我的麥克風占用了。有什麼想知道的都可以問我哦。你,你是什麼模型?不完全是呢,我是Elva,我們都是阿里集團研發的語言模型。但功能和特點可能有所不同。Elva,你確定嗎?你的語言模型不是簽文嗎?我就咱們倆在聊天呢。其他的就是根據需要調用不同的模型來提供幫助或者進行對話了。Elva,嗯,那你現在裡面都有哪幾個語言模型在裡面?我能回答問題,講故事,提建議,還能陪你聊天,解解悶。你想要聊些什麼?哦,好的,那你都會哪些模型?你,你會哪些模型的功能啊。對,我的響應速度挺快的。基本上是你說完我馬上就能接上話,不會讓你整太久。有才事可以隨時跟我說。說完話,它能夠實時,就能夠識別出來。而且它還支持打斷。那可能是當時的情況來特殊。你看它,你中間在說話的時候,你可以把它打斷。當然可以,你想說啥就直接告訴我吧。哎,我記得,我記得我剛才跟他說話的時候能打斷的。他這個好像剛才這個沒,沒生效。現在呢,現在可以打斷嗎?啊,對,應該。嗯,好的。那這個就是一個小模型。那其實這個模型是可以在部署在手機上的。那你Mac上也是完全可以的哦。而且呢,如果你感興趣的話呢,相當於我們過來給大家看一下這個模型它的文件。你其實,如果你可以的話,你可以把這個模型。嗯,sorry。不應該呀。你都可以把這個模型它的,RM模型換成那個大模型。那它相當於是,嗯,這是用了JM4的FoistTokenator。這個是,嗯,這個是TTS嗎?好像不是TTS。還真是用了千文三的Valba這個模型。然後呢,它其實就相當於把幾個小模型的一個功能集成在一起。然後通過工程的能力,然後把它們進行做了一個拼接。嗯,那如果感興趣的話呢,可以來嘗試一下。這個模型還是非常小哈。應該是可以在手機上去部署起來的。當然呢,那可以等等,應該會有一些第三方的一些工具,會把這些產品模型提升起來。能在本地運行。目前它這個模型沒有量化之前是7個G。那比如說Q4的話,應該是在3、4個G左右的一個大小。完全是可以在手機上跑起來的。好的,今天就介紹在這兒。謝謝大家。謝謝大家。", "segments": [{"id": 0, "seek": 0, "start": 0.0, "end": 24.2, "text": "各位朋友大家好,歡迎大家關注昆蓬特,今天給大家介紹一個小模型叫SoloX DuPRO 0.6B,那這個模型呢它是一個全商工的語音交互,什麼叫全商工就是能夠實時對話的一個模型,它的模型只有0.6B,0.6B什麼概念?就是你在手機上就能運行一個全本地的一個對話的一個模型,那它的效果還是非常不錯的,一會給大家演示一下,我本地沒有去跑它。", "tokens": [50365, 22073, 19828, 15248, 11, 21315, 6868, 14899, 26432, 1431, 228, 39198, 105, 17682, 11, 12074, 17798, 6868, 30312, 38193, 8990, 7322, 41908, 39823, 19855, 50, 7902, 55, 5153, 47, 7142, 1958, 13, 21, 33, 11, 4184, 6287, 41908, 39823, 6240, 11284, 1541, 8990, 11319, 45581, 23323, 1546, 31348, 18034, 28455, 1369, 240, 11, 7598, 19855, 11319, 45581, 23323, 5620, 8225, 31649, 10376, 6611, 2855, 11103, 1546, 8990, 41908, 39823, 11, 45224, 41908, 39823, 35244, 15, 13, 21, 33, 11, 15, 13, 21, 33, 7598, 26181, 33679, 30, 5620, 41410, 11389, 17543, 5708, 3111, 8225, 32772, 8082, 8990, 11319, 8802, 10928, 1546, 8990, 2855, 11103, 1546, 8990, 41908, 39823, 11, 4184, 45224, 43076, 9319, 25583, 14392, 30967, 1546, 11, 2257, 6236, 17798, 6868, 31382, 25696, 8861, 11, 1654, 8802, 10928, 6963, 6734, 32585, 11284, 1543, 51575], "temperature": 0, "avg_logprob": -0.20982873612555905, "compression_ratio": 1.396103896103896, "no_speech_prob": 8.039699114781307e-12}, {"id": 1, "seek": 2420, "start": 24.2, "end": 37.6, "text": "首先第一個呢,它是用來還是VD模型的這種,然後它後面呢是用一個叫它那個語音模型是用了千問3的一個0.6B的一個模型應該是。", "tokens": [50365, 36490, 40760, 6240, 11, 11284, 1541, 9254, 3763, 25583, 53, 35, 41908, 39823, 1546, 33005, 11, 10213, 11284, 5661, 8833, 6240, 1541, 9254, 8990, 19855, 11284, 20754, 31348, 18034, 41908, 39823, 1541, 9254, 2289, 20787, 11361, 18, 1546, 8990, 15, 13, 21, 33, 1546, 8990, 41908, 39823, 26087, 1541, 1543, 51035], "temperature": 0, "avg_logprob": -0.18755393241768453, "compression_ratio": 1.382636655948553, "no_speech_prob": 1.1221326819133015e-11}, {"id": 2, "seek": 2420, "start": 38.22, "end": 54.019999999999996, "text": "然後它安裝我們可以看一下它的安裝的方式,它目前是Github一個倉庫,然後在Github上可以去運行推理它,因為目前對於這種全商工的這種推理的框架不多,所以比如說NAMA.CPP啊,它可能也還沒有支持這種模型。", "tokens": [51066, 10213, 11284, 16206, 44453, 5884, 6723, 28324, 45224, 16206, 44453, 1546, 9249, 27584, 11, 11284, 39004, 1541, 38, 355, 836, 8990, 2122, 231, 6346, 104, 11, 10213, 3581, 38, 355, 836, 5708, 6723, 6734, 32772, 8082, 33597, 13876, 11284, 11, 11471, 39004, 2855, 19488, 33005, 11319, 45581, 23323, 1546, 33005, 33597, 13876, 1546, 21416, 228, 7360, 114, 33364, 11, 7239, 36757, 4622, 45, 38136, 13, 34, 17755, 4905, 11, 11284, 16657, 6404, 7824, 6963, 35488, 33005, 41908, 39823, 1543, 51856], "temperature": 0, "avg_logprob": -0.18755393241768453, "compression_ratio": 1.382636655948553, "no_speech_prob": 1.1221326819133015e-11}, {"id": 3, "seek": 5420, "start": 55.1, "end": 68.2, "text": "那前段時間,OpenBnB他們開玩那個叫NAMA.CPP-OMINI的一個模型,那它那個裡面是能實現這些功能,但是那個框架的可能也是OpenBnB他們剛出的,然後不太通用。", "tokens": [50410, 4184, 8945, 28427, 20788, 11, 45569, 33, 77, 33, 20486, 8949, 19912, 20754, 19855, 45, 38136, 13, 34, 17755, 12, 5251, 1464, 40, 1546, 8990, 41908, 39823, 11, 4184, 11284, 20754, 32399, 1541, 8225, 10376, 9581, 29869, 32311, 8225, 11, 11189, 20754, 21416, 228, 7360, 114, 1546, 16657, 22021, 45569, 33, 77, 33, 20486, 16940, 7781, 1546, 11, 10213, 1960, 9455, 19550, 9254, 1543, 51065], "temperature": 0, "avg_logprob": -0.230051025390625, "compression_ratio": 1.2811387900355873, "no_speech_prob": 1.7634872728766737e-11}, {"id": 4, "seek": 5420, "start": 68.2, "end": 80.30000000000001, "text": "那這個模型呢,它的信息很少,那我們可以看一下,它底下呢,說它致謝了一下這個千問的JM,千問的這個LM和JM4的,這應該是個TTS吧。", "tokens": [51065, 4184, 6287, 41908, 39823, 6240, 11, 45224, 17665, 26460, 4563, 15686, 11, 46714, 6723, 28324, 11, 11284, 22816, 4438, 6240, 11, 4622, 11284, 6784, 112, 6350, 2289, 8861, 6287, 20787, 11361, 1546, 41, 44, 11, 20787, 11361, 1546, 6287, 43, 44, 12565, 41, 44, 19, 1546, 11, 2664, 26087, 1541, 3338, 51, 7327, 6062, 1543, 51670], "temperature": 0, "avg_logprob": -0.230051025390625, "compression_ratio": 1.2811387900355873, "no_speech_prob": 1.7634872728766737e-11}, {"id": 5, "seek": 8030, "start": 80.3, "end": 91.3, "text": "這相當於它是一個縫合怪,那本身呢,基於這種全商工呢,它其實就是,第一個呢,是做語音識別,類似於SR,然後呢,然後識別完了之後呢,底層還有一個LM模型,模型進行推理。", "tokens": [50365, 2664, 15106, 13118, 19488, 11284, 1541, 8990, 21030, 104, 14245, 26798, 11, 4184, 8802, 19847, 6240, 11, 26008, 19488, 33005, 11319, 45581, 23323, 6240, 11, 11284, 14139, 5620, 11, 40760, 6240, 11, 1541, 10907, 31348, 18034, 43143, 16158, 11, 40092, 7384, 120, 19488, 50, 49, 11, 10213, 6240, 11, 10213, 43143, 16158, 41665, 26362, 6240, 11, 22816, 9636, 97, 15569, 8990, 43, 44, 41908, 39823, 11, 41908, 39823, 18214, 8082, 33597, 13876, 1543, 50915], "temperature": 0, "avg_logprob": -0.12112041080699247, "compression_ratio": 1.3859649122807018, "no_speech_prob": 1.6847464395786105e-11}, {"id": 6, "seek": 8030, "start": 91.6, "end": 99.92, "text": "推理完成之後呢,再用TTS給你讀出來,就是這麼一個情況,然後只不過它處理速度極快,導致你以為是跟一個,呃,完整的一個模型進行對話。", "tokens": [50930, 33597, 13876, 41509, 26362, 6240, 11, 8623, 9254, 51, 7327, 17798, 2166, 7422, 222, 29741, 11, 5620, 21269, 8990, 39391, 11, 10213, 14003, 28193, 11284, 39289, 13876, 31217, 13127, 13387, 113, 10251, 11, 25647, 6784, 112, 2166, 3588, 6344, 1541, 9678, 8990, 11, 3606, 225, 11, 14128, 27662, 1546, 8990, 41908, 39823, 18214, 8082, 2855, 11103, 1543, 51346], "temperature": 0, "avg_logprob": -0.12112041080699247, "compression_ratio": 1.3859649122807018, "no_speech_prob": 1.6847464395786105e-11}, {"id": 7, "seek": 9992, "start": 99.92, "end": 101.92, "text": "那這個是它的一個Demo,我們來演示一下。", "tokens": [50365, 4184, 41427, 45224, 8990, 35, 36221, 11, 5884, 3763, 31382, 25696, 8861, 1543, 50465], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 8, "seek": 9992, "start": 101.92, "end": 104.92, "text": "你好,你是誰?", "tokens": [50465, 26410, 11, 32526, 20757, 30, 50615], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 9, "seek": 9992, "start": 104.92, "end": 106.92, "text": "嗯,占用時間了,咱們繼續。", "tokens": [50615, 21206, 11, 5322, 254, 9254, 20788, 2289, 11, 8975, 109, 4623, 38459, 1543, 50715], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 10, "seek": 9992, "start": 106.92, "end": 112.92, "text": "嗯,抱歉,可能,我的麥克風被這個占用了,那我。", "tokens": [50715, 21206, 11, 38382, 4287, 231, 11, 16657, 11, 14200, 3994, 98, 24881, 22713, 23238, 6287, 5322, 254, 9254, 2289, 11, 4184, 1654, 1543, 51015], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 11, "seek": 9992, "start": 112.92, "end": 113.92, "text": "我聽著呢。", "tokens": [51015, 1654, 21524, 19382, 6240, 1543, 51065], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 12, "seek": 9992, "start": 113.92, "end": 116.92, "text": "哦,不是,還是喊過來,我以為是我的麥克風占用了。", "tokens": [51065, 14659, 11, 7296, 11, 25583, 5234, 232, 49950, 11, 1654, 3588, 6344, 1541, 14200, 3994, 98, 24881, 22713, 5322, 254, 9254, 2289, 1543, 51215], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 13, "seek": 9992, "start": 116.92, "end": 118.92, "text": "有什麼想知道的都可以問我哦。", "tokens": [51215, 2412, 7598, 7093, 7758, 1546, 7182, 6723, 11361, 1654, 14659, 1543, 51315], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 14, "seek": 9992, "start": 118.92, "end": 122.92, "text": "你,你是什麼模型?", "tokens": [51315, 2166, 11, 32526, 7598, 41908, 39823, 30, 51515], "temperature": 0, "avg_logprob": -0.26770093885518736, "compression_ratio": 1.3109243697478992, "no_speech_prob": 2.2356826459568602e-11}, {"id": 15, "seek": 12292, "start": 122.92, "end": 126.92, "text": "不完全是呢,我是Elva,我們都是阿里集團研發的語言模型。", "tokens": [50365, 1960, 37100, 1541, 6240, 11, 15914, 17356, 2757, 11, 5884, 22796, 13594, 15759, 26020, 37766, 23230, 242, 14637, 1546, 31348, 12009, 41908, 39823, 1543, 50565], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 16, "seek": 12292, "start": 126.92, "end": 128.92000000000002, "text": "但功能和特點可能有所不同。", "tokens": [50565, 8395, 32311, 8225, 12565, 17682, 8216, 16657, 2412, 5966, 47123, 1543, 50665], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 17, "seek": 12292, "start": 128.92000000000002, "end": 131.92000000000002, "text": "Elva,你確定嗎?你的語言模型不是簽文嗎?", "tokens": [50665, 17356, 2757, 11, 2166, 24293, 12088, 7434, 30, 18961, 31348, 12009, 41908, 39823, 7296, 24579, 121, 17174, 7434, 30, 50815], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 18, "seek": 12292, "start": 131.92000000000002, "end": 134.92000000000002, "text": "我就咱們倆在聊天呢。", "tokens": [50815, 22020, 8975, 109, 4623, 2122, 228, 3581, 40096, 6135, 6240, 1543, 50965], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 19, "seek": 12292, "start": 134.92000000000002, "end": 138.92000000000002, "text": "其他的就是根據需要調用不同的模型來提供幫助或者進行對話了。", "tokens": [50965, 9572, 31309, 5620, 31337, 36841, 35748, 32943, 9254, 47123, 1546, 41908, 39823, 3763, 20949, 3254, 249, 32187, 37618, 31148, 18214, 8082, 2855, 11103, 2289, 1543, 51165], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 20, "seek": 12292, "start": 138.92000000000002, "end": 142.92000000000002, "text": "Elva,嗯,那你現在裡面都有哪幾個語言模型在裡面?", "tokens": [51165, 17356, 2757, 11, 21206, 11, 4184, 2166, 12648, 32399, 48121, 17028, 23575, 3338, 31348, 12009, 41908, 39823, 3581, 32399, 30, 51365], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 21, "seek": 12292, "start": 142.92000000000002, "end": 146.92000000000002, "text": "我能回答問題,講故事,提建議,還能陪你聊天,解解悶。", "tokens": [51365, 1654, 8225, 8350, 28223, 17197, 11, 11932, 43045, 6973, 11, 20949, 34157, 24686, 11, 7824, 8225, 8842, 103, 2166, 40096, 6135, 11, 17278, 17278, 14696, 114, 1543, 51565], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 22, "seek": 12292, "start": 146.92000000000002, "end": 148.92000000000002, "text": "你想要聊些什麼?", "tokens": [51565, 38386, 4275, 40096, 13824, 7598, 30, 51665], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 23, "seek": 12292, "start": 149.92000000000002, "end": 152.92000000000002, "text": "哦,好的,那你都會哪些模型?", "tokens": [51715, 14659, 11, 20715, 11, 4184, 2166, 7182, 6236, 17028, 13824, 41908, 39823, 30, 51865], "temperature": 0, "avg_logprob": -0.12858078002929688, "compression_ratio": 1.3970588235294117, "no_speech_prob": 2.03724207642475e-11}, {"id": 24, "seek": 15292, "start": 152.92, "end": 154.92, "text": "你,你會哪些模型的功能啊。", "tokens": [50365, 2166, 11, 2166, 6236, 17028, 13824, 41908, 39823, 1546, 32311, 8225, 4905, 1543, 50465], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 25, "seek": 15292, "start": 154.92, "end": 156.92, "text": "對,我的響應速度挺快的。", "tokens": [50465, 2855, 11, 14200, 13665, 123, 20481, 31217, 13127, 41046, 10251, 1546, 1543, 50565], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 26, "seek": 15292, "start": 156.92, "end": 160.92, "text": "基本上是你說完我馬上就能接上話,不會讓你整太久。", "tokens": [50565, 37946, 5708, 1541, 42920, 14128, 1654, 29098, 5708, 3111, 8225, 14468, 5708, 11103, 11, 21121, 21195, 2166, 27662, 9455, 25320, 1543, 50765], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 27, "seek": 15292, "start": 160.92, "end": 162.92, "text": "有才事可以隨時跟我說。", "tokens": [50765, 2412, 18888, 6973, 6723, 48890, 6611, 9678, 34206, 1543, 50865], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 28, "seek": 15292, "start": 162.92, "end": 164.92, "text": "說完話,它能夠實時,就能夠識別出來。", "tokens": [50865, 4622, 14128, 11103, 11, 11284, 8225, 31649, 10376, 6611, 11, 3111, 8225, 31649, 43143, 16158, 29741, 1543, 50965], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 29, "seek": 15292, "start": 164.92, "end": 166.92, "text": "而且它還支持打斷。", "tokens": [50965, 22942, 11284, 7824, 35488, 12467, 4307, 115, 1543, 51065], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 30, "seek": 15292, "start": 166.92, "end": 169.92, "text": "那可能是當時的情況來特殊。", "tokens": [51065, 4184, 16657, 1541, 13118, 6611, 1546, 39391, 3763, 17682, 15976, 232, 1543, 51215], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 31, "seek": 15292, "start": 169.92, "end": 171.92, "text": "你看它,你中間在說話的時候,你可以把它打斷。", "tokens": [51215, 16529, 11284, 11, 2166, 5975, 11016, 3581, 4622, 11103, 20643, 11, 42766, 42061, 12467, 4307, 115, 1543, 51315], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 32, "seek": 15292, "start": 171.92, "end": 173.92, "text": "當然可以,你想說啥就直接告訴我吧。", "tokens": [51315, 21707, 6723, 11, 38386, 4622, 3284, 98, 3111, 43297, 31080, 1654, 6062, 1543, 51415], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 33, "seek": 15292, "start": 173.92, "end": 177.92, "text": "哎,我記得,我記得我剛才跟他說話的時候能打斷的。", "tokens": [51415, 26755, 11, 1654, 30872, 11, 1654, 30872, 1654, 16940, 18888, 9678, 5000, 4622, 11103, 20643, 8225, 12467, 4307, 115, 1546, 1543, 51615], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 34, "seek": 15292, "start": 177.92, "end": 179.92, "text": "他這個好像剛才這個沒,沒生效。", "tokens": [51615, 5000, 6287, 33242, 16940, 18888, 6287, 4913, 11, 4913, 8244, 43076, 1543, 51715], "temperature": 0, "avg_logprob": -0.31041137865801766, "compression_ratio": 1.4797687861271676, "no_speech_prob": 1.833947230189814e-11}, {"id": 35, "seek": 17992, "start": 179.92, "end": 182.92, "text": "現在呢,現在可以打斷嗎?", "tokens": [50365, 12648, 6240, 11, 12648, 6723, 12467, 4307, 115, 7434, 30, 50515], "temperature": 0, "avg_logprob": -0.2536663327898298, "compression_ratio": 0.8985507246376812, "no_speech_prob": 1.8240190607921036e-11}, {"id": 36, "seek": 17992, "start": 182.92, "end": 183.92, "text": "啊,對,應該。", "tokens": [50515, 4905, 11, 2855, 11, 26087, 1543, 50565], "temperature": 0, "avg_logprob": -0.2536663327898298, "compression_ratio": 0.8985507246376812, "no_speech_prob": 1.8240190607921036e-11}, {"id": 37, "seek": 17992, "start": 183.92, "end": 185.92, "text": "嗯,好的。", "tokens": [50565, 21206, 11, 20715, 1543, 50665], "temperature": 0, "avg_logprob": -0.2536663327898298, "compression_ratio": 0.8985507246376812, "no_speech_prob": 1.8240190607921036e-11}, {"id": 38, "seek": 18592, "start": 185.92, "end": 188.92, "text": "那這個就是一個小模型。", "tokens": [50365, 4184, 6287, 5620, 8990, 7322, 41908, 39823, 1543, 50515], "temperature": 0, "avg_logprob": -0.16774031775338308, "compression_ratio": 1.2328042328042328, "no_speech_prob": 2.1484685558403882e-11}, {"id": 39, "seek": 18592, "start": 188.92, "end": 190.92, "text": "那其實這個模型是可以在部署在手機上的。", "tokens": [50515, 4184, 14139, 6287, 41908, 39823, 1541, 6723, 3581, 13470, 16469, 110, 3581, 11389, 17543, 5708, 1546, 1543, 50615], "temperature": 0, "avg_logprob": -0.16774031775338308, "compression_ratio": 1.2328042328042328, "no_speech_prob": 2.1484685558403882e-11}, {"id": 40, "seek": 18592, "start": 190.92, "end": 192.92, "text": "那你Mac上也是完全可以的哦。", "tokens": [50615, 4184, 2166, 34355, 5708, 22021, 37100, 6723, 1546, 14659, 1543, 50715], "temperature": 0, "avg_logprob": -0.16774031775338308, "compression_ratio": 1.2328042328042328, "no_speech_prob": 2.1484685558403882e-11}, {"id": 41, "seek": 18592, "start": 192.92, "end": 197.92, "text": "而且呢,如果你感興趣的話呢,相當於我們過來給大家看一下這個模型它的文件。", "tokens": [50715, 22942, 6240, 11, 45669, 9709, 44089, 39835, 21358, 6240, 11, 15106, 13118, 19488, 5884, 49950, 17798, 6868, 28324, 6287, 41908, 39823, 45224, 17174, 20485, 1543, 50965], "temperature": 0, "avg_logprob": -0.16774031775338308, "compression_ratio": 1.2328042328042328, "no_speech_prob": 2.1484685558403882e-11}, {"id": 42, "seek": 19792, "start": 197.92, "end": 201.92, "text": "你其實,如果你可以的話,你可以把這個模型。", "tokens": [50365, 2166, 14139, 11, 13119, 42766, 21358, 11, 42766, 16075, 6287, 41908, 39823, 1543, 50565], "temperature": 0, "avg_logprob": -0.5211897398296156, "compression_ratio": 1.206896551724138, "no_speech_prob": 2.4226044781361367e-11}, {"id": 43, "seek": 19792, "start": 201.92, "end": 202.92, "text": "嗯,sorry。", "tokens": [50565, 21206, 11, 47545, 1543, 50615], "temperature": 0, "avg_logprob": -0.5211897398296156, "compression_ratio": 1.206896551724138, "no_speech_prob": 2.4226044781361367e-11}, {"id": 44, "seek": 19792, "start": 202.92, "end": 203.92, "text": "不應該呀。", "tokens": [50615, 1960, 26087, 15348, 1543, 50665], "temperature": 0, "avg_logprob": -0.5211897398296156, "compression_ratio": 1.206896551724138, "no_speech_prob": 2.4226044781361367e-11}, {"id": 45, "seek": 19792, "start": 203.92, "end": 211.92, "text": "你都可以把這個模型它的,RM模型換成那個大模型。", "tokens": [50665, 2166, 7182, 6723, 16075, 6287, 41908, 39823, 45224, 11, 49, 44, 41908, 39823, 36338, 11336, 20754, 3582, 41908, 39823, 1543, 51065], "temperature": 0, "avg_logprob": -0.5211897398296156, "compression_ratio": 1.206896551724138, "no_speech_prob": 2.4226044781361367e-11}, {"id": 46, "seek": 19792, "start": 211.92, "end": 215.92, "text": "那它相當於是,嗯,這是用了JM4的FoistTokenator。", "tokens": [51065, 4184, 11284, 15106, 13118, 19488, 1541, 11, 21206, 11, 16600, 9254, 2289, 41, 44, 19, 1546, 37, 78, 468, 51, 8406, 1639, 1543, 51265], "temperature": 0, "avg_logprob": -0.5211897398296156, "compression_ratio": 1.206896551724138, "no_speech_prob": 2.4226044781361367e-11}, {"id": 47, "seek": 21592, "start": 215.92, "end": 220.92, "text": "這個是,嗯,這個是TTS嗎?好像不是TTS。", "tokens": [50365, 41427, 11, 21206, 11, 41427, 51, 7327, 7434, 30, 33242, 7296, 51, 7327, 1543, 50615], "temperature": 0, "avg_logprob": -0.21874362721162685, "compression_ratio": 1.2412060301507537, "no_speech_prob": 2.1996317961514578e-11}, {"id": 48, "seek": 21592, "start": 220.92, "end": 224.92, "text": "還真是用了千文三的Valba這個模型。", "tokens": [50615, 7824, 6303, 1541, 9254, 2289, 20787, 17174, 10960, 1546, 53, 304, 4231, 6287, 41908, 39823, 1543, 50815], "temperature": 0, "avg_logprob": -0.21874362721162685, "compression_ratio": 1.2412060301507537, "no_speech_prob": 2.1996317961514578e-11}, {"id": 49, "seek": 21592, "start": 224.92, "end": 228.92, "text": "然後呢,它其實就相當於把幾個小模型的一個功能集成在一起。", "tokens": [50815, 10213, 6240, 11, 11284, 14139, 3111, 15106, 13118, 19488, 16075, 23575, 3338, 7322, 41908, 39823, 1546, 8990, 32311, 8225, 26020, 11336, 3581, 29567, 1543, 51015], "temperature": 0, "avg_logprob": -0.21874362721162685, "compression_ratio": 1.2412060301507537, "no_speech_prob": 2.1996317961514578e-11}, {"id": 50, "seek": 21592, "start": 228.92, "end": 232.92, "text": "然後通過工程的能力,然後把它們進行做了一個拼接。", "tokens": [51015, 10213, 19550, 8816, 23323, 29649, 1546, 8225, 13486, 11, 10213, 42061, 4623, 18214, 8082, 10907, 2289, 8990, 6852, 120, 14468, 1543, 51215], "temperature": 0, "avg_logprob": -0.21874362721162685, "compression_ratio": 1.2412060301507537, "no_speech_prob": 2.1996317961514578e-11}, {"id": 51, "seek": 23292, "start": 232.92, "end": 243.92, "text": "嗯,那如果感興趣的話呢,可以來嘗試一下。這個模型還是非常小哈。", "tokens": [50365, 21206, 11, 4184, 13119, 9709, 44089, 39835, 21358, 6240, 11, 6723, 3763, 10756, 245, 22099, 8861, 1543, 6287, 41908, 39823, 25583, 14392, 7322, 10595, 1543, 50915], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 52, "seek": 23292, "start": 243.92, "end": 246.92, "text": "應該是可以在手機上去部署起來的。", "tokens": [50915, 26087, 1541, 6723, 3581, 11389, 17543, 5708, 6734, 13470, 16469, 110, 21670, 1546, 1543, 51065], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 53, "seek": 23292, "start": 246.92, "end": 251.92, "text": "當然呢,那可以等等,應該會有一些第三方的一些工具,會把這些產品模型提升起來。", "tokens": [51065, 21707, 6240, 11, 4184, 6723, 36000, 11, 26087, 6236, 32241, 13824, 35878, 9249, 1546, 38515, 23323, 39806, 11, 6236, 16075, 29869, 33299, 30246, 41908, 39823, 20949, 41670, 21670, 1543, 51315], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 54, "seek": 23292, "start": 251.92, "end": 254.92, "text": "能在本地運行。目前它這個模型沒有量化之前是7個G。", "tokens": [51315, 8225, 3581, 8802, 10928, 32772, 8082, 1543, 39004, 11284, 6287, 41908, 39823, 6963, 26748, 23756, 32442, 1541, 22, 3338, 38, 1543, 51465], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 55, "seek": 23292, "start": 254.92, "end": 257.91999999999996, "text": "那比如說Q4的話,應該是在3、4個G左右的一個大小。", "tokens": [51465, 4184, 36757, 4622, 48, 19, 21358, 11, 26087, 1541, 3581, 18, 1231, 19, 3338, 38, 29457, 1546, 8990, 3582, 7322, 1543, 51615], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 56, "seek": 23292, "start": 257.91999999999996, "end": 258.91999999999996, "text": "完全是可以在手機上跑起來的。", "tokens": [51615, 37100, 1541, 6723, 3581, 11389, 17543, 5708, 32585, 21670, 1546, 1543, 51665], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 57, "seek": 23292, "start": 258.91999999999996, "end": 260.91999999999996, "text": "好的,今天就介紹在這兒。謝謝大家。", "tokens": [51665, 20715, 11, 12074, 3111, 30312, 38193, 3581, 2664, 35125, 1543, 11003, 6868, 1543, 51765], "temperature": 0, "avg_logprob": -0.1409039815266927, "compression_ratio": 1.4420731707317074, "no_speech_prob": 1.6966952148811387e-11}, {"id": 58, "seek": 26092, "start": 260.92, "end": 262.92, "text": "謝謝大家。", "tokens": [50365, 11003, 6868, 1543, 50465], "temperature": 0, "avg_logprob": -0.19276841481526694, "compression_ratio": 0.625, "no_speech_prob": 1.6362652552337487e-11}], "language": "zh"}