data:image/png;base64,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5:["$","main",null,{"className":"journey-reader-page min-h-screen bg-background text-primary","children":["$","$L16",null,{"title":"Fast Transformer Decoding: One Write-Head Is All You Need — MQA","pdfUrl":"/journey/papers/fast-transformer-decoding-one-write-head-is-all-you-need-mqa.pdf","annotations":[{"id":"01M1K4Y0VN9RXR477SHQQ254RV","type":"text","color":"#57c785","content":{"text":"We propose a variant called multi-query attention","image":null},"position":{"page":1,"boundingRect":{"page":1,"x":0.4656191867154232,"y":0.39318176363855156,"width":0.32753210561845725,"height":0.013160035496040928},"rects":[{"page":1,"x":0.4656191867154232,"y":0.39318176363855156,"width":0.32753210561845725,"height":0.013160035496040928}]},"comments":[],"commentCount":0,"createdAt":"2026-09-03T08:07:29.397Z","updatedAt":"2026-09-03T08:07:29.397Z"},{"id":"01M1K50TJZDFBT857KX3VRQ0H3","type":"text","color":"#f2c94c","content":{"text":". One major challenge with Transformer is the speed of incremental inference","image":null},"position":{"page":1,"boundingRect":{"page":1,"x":0.27863776895321374,"y":0.5469896784985301,"width":0.5517514880622338,"height":0.01425670512071102},"rects":[{"page":1,"x":0.27863776895321374,"y":0.5469896784985301,"width":0.5517514880622338,"height":0.01425670512071102}]},"comments":[{"id":"01M1K50WX2G7QG9694ATWQ4EVH","parentId":null,"isAi":false,"prompt":null,"html":"

Incremental inference = inferenza autoregressiva in cui l'output viene costruito incrementalmente, un token alla volta, usando a ogni passo tutti i token precedentemente disponibili.

","text":"Incremental inference = inferenza autoregressiva in cui l'output viene costruito incrementalmente, un token alla volta, usando a ogni passo tutti i token precedentemente disponibili.","createdAt":"2026-09-03T08:09:03.651Z","updatedAt":"2026-09-03T08:09:03.651Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:09:01.279Z","updatedAt":"2026-09-03T08:09:01.279Z"},{"id":"01M1K55DZ2AQAEP3BH62D7PFHX","type":"text","color":"#f2c94c","content":{"text":"l o g i t s = t f . einsum ( \" k , mk−>m\" , q , K)","image":null},"position":{"page":2,"boundingRect":{"page":2,"x":0.16461142654364797,"y":0.23386479746089434,"width":0.33665341769151436,"height":0.01425670512071102},"rects":[{"page":2,"x":0.16461142654364797,"y":0.23386479746089434,"width":0.33665341769151436,"height":0.01425670512071102}]},"comments":[{"id":"01M1K55H5C5VRR39NMX3KRE7MX","parentId":null,"isAi":false,"prompt":null,"html":"

score di compatibilità per ciascun token precedente.

","text":"score di compatibilità per ciascun token precedente.","createdAt":"2026-09-03T08:11:35.468Z","updatedAt":"2026-09-03T08:11:35.468Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:11:32.194Z","updatedAt":"2026-09-03T08:11:32.194Z"},{"id":"01M1K5CHG0HN7ZVHC6YCJW8W0M","type":"text","color":"#f2c94c","content":{"text":"l o g i t s = t f . einsum ( \" bhnk , bhmk−>bhnm \" , Q, K)","image":null},"position":{"page":3,"boundingRect":{"page":3,"x":0.16449791003184713,"y":0.45416346153846154,"width":0.41441884617896596,"height":0.014769230769230769},"rects":[{"page":3,"x":0.16449791003184713,"y":0.45416346153846154,"width":0.41441884617896596,"height":0.014769230769230769}]},"comments":[{"id":"01M1K5CMW71F9WFPXQ4AM7DHCA","parentId":null,"isAi":false,"prompt":null,"html":"

Per ogni sequenza, per ogni head, per ogni Query, calcola contemporaneamente la compatibilità con tutte le Key.

","text":"Per ogni sequenza, per ogni head, per ogni Query, calcola contemporaneamente la compatibilità con tutte le Key.","createdAt":"2026-09-03T08:15:28.647Z","updatedAt":"2026-09-03T08:15:28.647Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:15:25.184Z","updatedAt":"2026-09-03T08:15:25.184Z"},{"id":"01M1K5FK3PR3Y1N8HKMN6MBCD9","type":"text","color":"#f2c94c","content":{"text":"O( 1 k + 1 bn )","image":null},"position":{"page":3,"boundingRect":{"page":3,"x":0.7407941878980892,"y":0.7526923076923077,"width":0.06998351273263335,"height":0.018423076923076924},"rects":[{"page":3,"x":0.7407941878980892,"y":0.7536634615384615,"width":0.06998351273263335,"height":0.014769230769230769},{"page":3,"x":0.7621914808917197,"y":0.7526923076923077,"width":0.006344546178343949,"height":0.009230769230769232},{"page":3,"x":0.7928941082802548,"y":0.7526923076923077,"width":0.006344546178343949,"height":0.009230769230769232},{"page":3,"x":0.7617933917197452,"y":0.7618846153846154,"width":0.010076632165605096,"height":0.009230769230769232},{"page":3,"x":0.7891993431528662,"y":0.7618846153846154,"width":0.016869028662420384,"height":0.009230769230769232}]},"comments":[{"id":"01M1K5FP1JRB0KCX4RYJDKX162","parentId":null,"isAi":false,"prompt":null,"html":"

carico relativamente pochi dati dalla memoria → faccio tantissimo calcolo con quei dati.

","text":"carico relativamente pochi dati dalla memoria → faccio tantissimo calcolo con quei dati.","createdAt":"2026-09-03T08:17:08.146Z","updatedAt":"2026-09-03T08:17:08.146Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:17:05.142Z","updatedAt":"2026-09-03T08:17:05.142Z"},{"id":"01M1K5KHBHEH4XGD89165B8R9J","type":"text","color":"#57c785","content":{"text":"bhnk , bhmk−>bhnm","image":null},"position":{"page":3,"boundingRect":{"page":3,"x":0.35843523097901037,"y":0.45403095156889495,"width":0.1431201689030781,"height":0.01432469304229195},"rects":[{"page":3,"x":0.35843523097901037,"y":0.45403095156889495,"width":0.1431201689030781,"height":0.01432469304229195}]},"comments":[{"id":"01M1K5KMNFQVDTY9CF6JQF9NSJ","parentId":null,"isAi":false,"prompt":null,"html":"

\"Prendi Q [batch, head, query, dimensione-vettore] e K [batch, head, key, dimensione-vettore]; fai il prodotto scalare sulla dimensione ke restituiscimi, per ogni batch, head, query e key, il relativo score.\"

","text":"\"Prendi Q [batch, head, query, dimensione-vettore] e K [batch, head, key, dimensione-vettore]; fai il prodotto scalare sulla dimensione ke restituiscimi, per ogni batch, head, query e key, il relativo score.\"","createdAt":"2026-09-03T08:19:17.807Z","updatedAt":"2026-09-03T08:19:17.807Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:19:14.417Z","updatedAt":"2026-09-03T08:19:14.417Z"},{"id":"01M1K5MKJQ3J4K8KYBQ9DMTSS7","type":"text","color":"#57c785","content":{"text":"when generating from the trained model, the output of the self-attention layer at a particular position affects the token that is generated at the next position","image":null},"position":{"page":4,"boundingRect":{"page":4,"x":0.11759432921447485,"y":0.09203402114597545,"width":0.764966057342233,"height":0.044519525920873125},"rects":[{"page":4,"x":0.6400716433142101,"y":0.09203402114597545,"width":0.24221174167723963,"height":0.01432469304229195},{"page":4,"x":0.11759432921447485,"y":0.10791481923601637,"width":0.764966057342233,"height":0.01330150068212824},{"page":4,"x":0.11759432921447485,"y":0.12222885402455662,"width":0.12440510924053949,"height":0.01432469304229195}]},"comments":[{"id":"01M1K5PDNKPTSMPKHFP0DC9A0S","parentId":null,"isAi":false,"prompt":null,"html":"

Le tecniche sopra citate possono essere utili in fase di training ma quando andiamo ad inferenziare il modello finale la generazione dei token sarà sempre progressiva e non distribuibile.

","text":"Le tecniche sopra citate possono essere utili in fase di training ma quando andiamo ad inferenziare il modello finale la generazione dei token sarà sempre progressiva e non distribuibile.","createdAt":"2026-09-03T08:20:48.947Z","updatedAt":"2026-09-03T08:20:48.947Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:19:49.463Z","updatedAt":"2026-09-03T08:19:49.463Z"},{"id":"01M1K5TR1CD8XT401HQVP16MTB","type":"text","color":"#57c785","content":{"text":"Multi-head attention consists of multiple attention layers (heads) in parallel with different linear transformations on the queries, keys, values and output","image":null},"position":{"page":5,"boundingRect":{"page":5,"x":0.11759334415584416,"y":0.13456541218637993,"width":0.7649333437210112,"height":0.030152329749103944},"rects":[{"page":5,"x":0.17178803339517626,"y":0.13456541218637993,"width":0.710738654481679,"height":0.015053763440860216},{"page":5,"x":0.11759334415584416,"y":0.14966397849462365,"width":0.40511705614419063,"height":0.015053763440860216}]},"comments":[{"id":"01M1K5WE7HTQA8QP8P1ZP6N845","parentId":null,"isAi":false,"prompt":null,"html":"

Questa è la referenza all’approccio del paper originale che citava il multi head mirato proprio al concedere al modello la possibilità di “cocentrarsi” su differenti caratteristiche del testo.

","text":"Questa è la referenza all’approccio del paper originale che citava il multi head mirato proprio al concedere al modello la possibilità di “cocentrarsi” su differenti caratteristiche del testo.","createdAt":"2026-09-03T08:24:06.129Z","updatedAt":"2026-09-03T08:24:06.129Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:23:10.636Z","updatedAt":"2026-09-03T08:23:10.636Z"},{"id":"01M1K5X4F7ME83HFF6SJV4BYE4","type":"text","color":"#f2c94c","content":{"text":"Multi-query attention is identical except that the different heads share a single set of keys and value","image":null},"position":{"page":5,"boundingRect":{"page":5,"x":0.11759334415584416,"y":0.14966397849462365,"width":0.7645324254080184,"height":0.030253136200716846},"rects":[{"page":5,"x":0.533963629841141,"y":0.14966397849462365,"width":0.3481621397227215,"height":0.015053763440860216},{"page":5,"x":0.11759334415584416,"y":0.1648633512544803,"width":0.3678108172867869,"height":0.015053763440860216}]},"comments":[{"id":"01M1K5Z8Q2EPJVTTQHTNTV0ZE9","parentId":null,"isAi":false,"prompt":null,"html":"

Il nuovo sistema proposto non cambia l’idea di avere “N” teste ma si assicura che queste condividano gli stessi vettori di referenza per K e V andando quindi tutti ad operare su quest’utlimi

","text":"Il nuovo sistema proposto non cambia l’idea di avere “N” teste ma si assicura che queste condividano gli stessi vettori di referenza per K e V andando quindi tutti ad operare su quest’utlimi","createdAt":"2026-09-03T08:25:38.786Z","updatedAt":"2026-09-03T08:28:08.171Z","replies":[]},{"id":"01M1K6507Z9Q7J85NFQEH3CTYQ","parentId":null,"isAi":false,"prompt":null,"html":"

Quindi la backpropagation deve far sì che quelle K e V condivise diventino rappresentazioni utili contemporaneamente alle diverse Query.

","text":"Quindi la backpropagation deve far sì che quelle K e V condivise diventino rappresentazioni utili contemporaneamente alle diverse Query.","createdAt":"2026-09-03T08:28:46.719Z","updatedAt":"2026-09-03T08:28:46.719Z","replies":[]}],"commentCount":2,"createdAt":"2026-09-03T08:24:28.903Z","updatedAt":"2026-09-03T08:24:28.903Z"},{"id":"01M1K60SBP3J0NZRHXNGYN9DTB","type":"area","color":"#f2c94c","content":{"text":null,"image":"$17"},"position":{"page":4,"boundingRect":{"page":4,"x":0.12193692835365853,"y":0.15812712202050996,"width":0.8112185594512196,"height":0.431313011017184},"rects":[]},"comments":[{"id":"01M1K60WX183Y1KTK1JW0P9P9X","parentId":null,"isAi":false,"prompt":null,"html":"

Nell’incremental self-attention cambia soprattutto una cosa: non ricalcoli K e V di tutti i token precedenti. Li tieni in cache e aggiungi solo quelli del nuovo token.

","text":"Nell’incremental self-attention cambia soprattutto una cosa: non ricalcoli K e V di tutti i token precedenti. Li tieni in cache e aggiungi solo quelli del nuovo token.","createdAt":"2026-09-03T08:26:32.225Z","updatedAt":"2026-09-03T08:26:32.225Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:26:28.598Z","updatedAt":"2026-09-03T08:26:28.598Z"},{"id":"01M1K6F86JZ0VCN04AGJKDMH7H","type":"text","color":"#f2c94c","content":{"text":"we find that the amortized inference time is 1.7μs per token for the encoder and a much larger 46μs per token for the decoder","image":null},"position":{"page":7,"boundingRect":{"page":7,"x":0.1170998831775701,"y":0.7376828937861272,"width":0.7659228743793809,"height":0.02955653901734104},"rects":[{"page":7,"x":0.19409915531907126,"y":0.7376828937861272,"width":0.6889236022378796,"height":0.014450867052023121},{"page":7,"x":0.1170998831775701,"y":0.752788565751445,"width":0.22745258652161215,"height":0.014450867052023121}]},"comments":[{"id":"01M1K6FAQ5WPHMKQD0TR6XQV24","parentId":null,"isAi":false,"prompt":null,"html":"

il decoder incrementale diventa circa 12× più veloce.

","text":"il decoder incrementale diventa circa 12× più veloce.","createdAt":"2026-09-03T08:34:25.125Z","updatedAt":"2026-09-03T08:34:25.125Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:34:22.547Z","updatedAt":"2026-09-03T08:34:22.547Z"},{"id":"01M1K6J0R290SR6Q7PWRKG81DD","type":"area","color":"#f2c94c","content":{"text":null,"image":"$18"},"position":{"page":8,"boundingRect":{"page":8,"x":0.10224518107476635,"y":0.6956878838511561,"width":0.8147415303738318,"height":0.0652518740968208},"rects":[]},"comments":[{"id":"01M1K6J2BM47Z8CVBAH12AENH4","parentId":null,"isAi":false,"prompt":null,"html":"

se costringi tutte le head a condividere K e V fin dal training, il modello riesce ad adattarsi a questo vincolo perdendo sorprendentemente poca qualità.

","text":"se costringi tutte le head a condividere K e V fin dal training, il modello riesce ad adattarsi a questo vincolo perdendo sorprendentemente poca qualità.","createdAt":"2026-09-03T08:35:54.868Z","updatedAt":"2026-09-03T08:36:51.067Z","replies":[]},{"id":"01M1K6JX5ZQJB0B55VPCZ8KP5N","parentId":null,"isAi":false,"prompt":null,"html":"

Tutte hanno consultato le stesse Key, ma facendo domande diverse tramite Query diverse hanno ottenuto distribuzioni di attenzione completamente diverse.

","text":"Tutte hanno consultato le stesse Key, ma facendo domande diverse tramite Query diverse hanno ottenuto distribuzioni di attenzione completamente diverse.","createdAt":"2026-09-03T08:36:22.335Z","updatedAt":"2026-09-03T08:36:22.335Z","replies":[]},{"id":"01M1K6NS1EGC674K672RQ9SYXY","parentId":null,"isAi":false,"prompt":"Explanation","html":null,"text":"**Page 8 Annotation: Key Performance Results** \n- **Inference speed:** MQA reduces decoder latency from **46 μs/token** (MHA) to just **3.8 μs/token**—a **12x speedup**. \n- **Quality retention:** \n - BLEU (WMT14): **28.5** (MQA) vs 28.3 (MHA baseline) \n - Perplexity (Billion-Word LM): **54.8** (MQA) vs 54.6 (MHA) \n- **Takeaway:** Validates MQA's core claim—drastically faster Transformer decoding (*especially for autoregressive tasks*) with minimal quality loss, by sharing keys/values across heads.","createdAt":"2026-09-03T08:37:56.398Z","updatedAt":"2026-09-03T08:37:56.398Z","replies":[]}],"commentCount":3,"createdAt":"2026-09-03T08:35:53.218Z","updatedAt":"2026-09-03T08:35:53.218Z"},{"id":"01M1K6M4GNEDYY8D4049K8RAZX","type":"text","color":"#57c785","content":{"text":"Conclusion","image":null},"position":{"page":8,"boundingRect":{"page":8,"x":0.15728679906542056,"y":0.6708363439306358,"width":0.12612102187682536,"height":0.020953757225433526},"rects":[{"page":8,"x":0.15728679906542056,"y":0.6708363439306358,"width":0.12612102187682536,"height":0.020953757225433526}]},"comments":[{"id":"01M1K6M6AFSBJZQAS7PZ2SXGH3","parentId":null,"isAi":false,"prompt":null,"html":"

Hanno scoperto empiricamente che il modello non ha bisogno di tutta quella libertà separata su K e V per mantenere quasi la stessa qualità: può imparare K/V condivisi sufficientemente ricchi, lasciando alle diverse Query il compito di estrarne informazioni differenti.

","text":"Hanno scoperto empiricamente che il modello non ha bisogno di tutta quella libertà separata su K e V per mantenere quasi la stessa qualità: può imparare K/V condivisi sufficientemente ricchi, lasciando alle diverse Query il compito di estrarne informazioni differenti.","createdAt":"2026-09-03T08:37:04.463Z","updatedAt":"2026-09-03T08:37:04.463Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:37:02.613Z","updatedAt":"2026-09-03T08:37:02.613Z"},{"id":"01M1K6VMB08RGQ0ZVTB76QS500","type":"text","color":"#57c785","content":{"text":"raining a","image":null},"position":{"page":7,"boundingRect":{"page":7,"x":0.6357171143724325,"y":0.6317490925589837,"width":0.06262794458810153,"height":0.014519056261343012},"rects":[{"page":7,"x":0.6357171143724325,"y":0.6317490925589837,"width":0.06262794458810153,"height":0.014519056261343012}]},"comments":[{"id":"01M1K6VNY0JM3RZANRJBN5NK82","parentId":null,"isAi":false,"prompt":null,"html":"

La backpropagation porta W_K e W_V condivise ad apprendere rappresentazioni che siano complessivamente utili alle diverse query prodotte da tutte le attention head.

","text":"La backpropagation porta W_K e W_V condivise ad apprendere rappresentazioni che siano complessivamente utili alle diverse query prodotte da tutte le attention head.","createdAt":"2026-09-03T08:41:09.824Z","updatedAt":"2026-09-03T08:41:09.824Z","replies":[]}],"commentCount":1,"createdAt":"2026-09-03T08:41:08.192Z","updatedAt":"2026-09-03T08:41:08.192Z"}],"backHref":"/en/journey","labels":{"back":"Back to The Journey","notes":"Notes & annotations","analysis":"Analysis","openAnalysis":"Open analysis","closeAnalysis":"Close analysis","resizeAnalysis":"Resize","page":"Page","loadError":"This PDF could not be loaded. Check that the exported paper is available.","source":"Source"},"analysis":"$L19","year":2019,"sourceUrl":"https://arxiv.org/pdf/1911.02150"}]}]

Work10 works

Projects8 projects

The Journey8 papers

About23 years

Work10 works

Projects8 projects

The Journey8 papers

About23 years

Work10 works

Projects8 projects

The Journey8 papers

About23 years