{"id":46237,"date":"2024-09-04T14:05:16","date_gmt":"2024-09-04T14:05:16","guid":{"rendered":"https:\/\/fabricegrinda.com\/?p=46237"},"modified":"2024-09-05T14:50:37","modified_gmt":"2024-09-05T14:50:37","slug":"fabrice-ai-hanh-trinh-ky-thuat","status":"publish","type":"post","link":"https:\/\/grinda.org\/vi\/fabrice-ai-hanh-trinh-ky-thuat\/","title":{"rendered":"Fabrice AI: H\u00e0nh tr\u00ecnh k\u1ef9 thu\u1eadt"},"content":{"rendered":"\n

Nh\u01b0 t\u00f4i \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc<\/a> , vi\u1ec7c ph\u00e1t tri\u1ec3n Fabrice AI<\/a> ph\u1ee9c t\u1ea1p h\u01a1n d\u1ef1 ki\u1ebfn r\u1ea5t nhi\u1ec1u, bu\u1ed9c t\u00f4i ph\u1ea3i kh\u00e1m ph\u00e1 nhi\u1ec1u c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c nhau.<\/p>\n\n

C\u00e1ch ti\u1ebfp c\u1eadn ban \u0111\u1ea7u: Ch\u1ec9 s\u1ed1 Llama – T\u00ecm ki\u1ebfm Vector<\/strong><\/p>\n\n

L\u1ea7n \u0111\u1ea7u ti\u00ean t\u00f4i th\u1eed nghi\u1ec7m n\u00e2ng cao kh\u1ea3 n\u0103ng truy xu\u1ea5t c\u1ee7a Fabrice AI li\u00ean quan \u0111\u1ebfn vi\u1ec7c s\u1eed d\u1ee5ng Llama Index \u0111\u1ec3 t\u00ecm ki\u1ebfm vector. Kh\u00e1i ni\u1ec7m n\u00e0y r\u1ea5t \u0111\u01a1n gi\u1ea3n: l\u1ea5y n\u1ed9i dung t\u1eeb blog c\u1ee7a t\u00f4i, chuy\u1ec3n \u0111\u1ed5i th\u00e0nh t\u00e0i li\u1ec7u Langchain, sau \u0111\u00f3 chuy\u1ec3n \u0111\u1ed5i ch\u00fang th\u00e0nh t\u00e0i li\u1ec7u Llama. C\u00e1c t\u00e0i li\u1ec7u Llama n\u00e0y sau \u0111\u00f3 s\u1ebd \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong m\u1ed9t vector index, cho ph\u00e9p t\u00f4i truy v\u1ea5n ch\u1ec9 m\u1ee5c n\u00e0y \u0111\u1ec3 t\u00ecm th\u00f4ng tin c\u00f3 li\u00ean quan. <\/p>\n\n

Tuy nhi\u00ean, khi t\u00f4i b\u1eaft \u0111\u1ea7u th\u1eed nghi\u1ec7m h\u1ec7 th\u1ed1ng, t\u00f4i nh\u1eadn ra r\u1eb1ng c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y kh\u00f4ng mang l\u1ea1i k\u1ebft qu\u1ea3 nh\u01b0 mong \u0111\u1ee3i. C\u1ee5 th\u1ec3, khi t\u00f4i truy v\u1ea5n h\u1ec7 th\u1ed1ng b\u1eb1ng nh\u1eefng c\u00e2u h\u1ecfi n\u1eb7ng v\u1ec1 ng\u1eef c\u1ea3nh nh\u01b0 “Nh\u1eefng sai l\u1ea7m l\u1edbn nh\u1ea5t m\u00e0 ng\u01b0\u1eddi s\u00e1ng l\u1eadp th\u1ecb tr\u01b0\u1eddng m\u1eafc ph\u1ea3i l\u00e0 g\u00ec?”, AI \u0111\u00e3 kh\u00f4ng \u0111\u01b0a ra \u0111\u01b0\u1ee3c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 \u00fd ngh\u0129a. Thay v\u00ec truy xu\u1ea5t n\u1ed9i dung s\u1eafc th\u00e1i m\u00e0 t\u00f4i bi\u1ebft l\u00e0 \u0111\u00e3 \u0111\u01b0\u1ee3c nh\u00fang trong d\u1eef li\u1ec7u, n\u00f3 tr\u1ea3 v\u1ec1 nh\u1eefng ph\u1ea3n h\u1ed3i kh\u00f4ng li\u00ean quan ho\u1eb7c kh\u00f4ng \u0111\u1ea7y \u0111\u1ee7. <\/p>\n\n

Th\u1ea5t b\u1ea1i ban \u0111\u1ea7u n\u00e0y khi\u1ebfn t\u00f4i ph\u1ea3i xem x\u00e9t l\u1ea1i c\u00e1ch ti\u1ebfp c\u1eadn c\u1ee7a m\u00ecnh. T\u00f4i nh\u1eadn ra r\u1eb1ng ch\u1ec9 l\u01b0u tr\u1eef n\u1ed9i dung trong m\u1ed9t ch\u1ec9 m\u1ee5c vect\u01a1 l\u00e0 kh\u00f4ng \u0111\u1ee7; c\u01a1 ch\u1ebf truy xu\u1ea5t c\u1ea7n ph\u1ea3i hi\u1ec3u b\u1ed1i c\u1ea3nh v\u00e0 s\u1eafc th\u00e1i c\u1ee7a c\u00e1c c\u00e2u h\u1ecfi \u0111\u01b0\u1ee3c h\u1ecfi. Nh\u1eadn th\u1ee9c n\u00e0y l\u00e0 b\u00e0i h\u1ecdc \u0111\u1ea7u ti\u00ean trong s\u1ed1 nhi\u1ec1u b\u00e0i h\u1ecdc s\u1ebd \u0111\u1ecbnh h\u00ecnh s\u1ef1 ph\u00e1t tri\u1ec3n c\u1ee7a Fabrice AI. <\/p>\n\n

L\u01b0u tr\u1eef ki\u1ebfn th\u1ee9c: L\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t t\u00e0i li\u1ec7u MongoDB<\/strong><\/p>\n\n

V\u1edbi nh\u1eefng h\u1ea1n ch\u1ebf c\u1ee7a ph\u01b0\u01a1ng ph\u00e1p Llama Index, ti\u1ebfp theo t\u00f4i kh\u00e1m ph\u00e1 c\u00e1ch l\u01b0u tr\u1eef c\u00e1c t\u00e0i li\u1ec7u Llama trong MongoDB. S\u01a1 \u0111\u1ed3 linh ho\u1ea1t v\u00e0 c\u1ea5u tr\u00fac h\u01b0\u1edbng t\u00e0i li\u1ec7u c\u1ee7a MongoDB c\u00f3 v\u1ebb l\u00e0 gi\u1ea3i ph\u00e1p \u0111\u1ea7y h\u1ee9a h\u1eb9n \u0111\u1ec3 qu\u1ea3n l\u00fd nhi\u1ec1u lo\u1ea1i n\u1ed9i dung kh\u00e1c nhau m\u00e0 t\u00f4i \u0111\u00e3 t\u00edch l\u0169y trong nhi\u1ec1u n\u0103m. <\/p>\n\n

K\u1ebf ho\u1ea1ch l\u00e0 t\u1ea1o ra tr\u1ea3i nghi\u1ec7m t\u00ecm ki\u1ebfm n\u0103ng \u0111\u1ed9ng v\u00e0 ph\u1ea3n h\u1ed3i h\u01a1n. Tuy nhi\u00ean, c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y nhanh ch\u00f3ng g\u1eb7p ph\u1ea3i v\u1ea5n \u0111\u1ec1. Ch\u1ee9c n\u0103ng t\u00ecm ki\u1ebfm m\u00e0 t\u00f4i d\u1ef1 \u0111o\u00e1n s\u1ebd m\u1ea1nh m\u1ebd h\u01a1n, \u0111\u00e3 kh\u00f4ng ho\u1ea1t \u0111\u1ed9ng nh\u01b0 mong \u0111\u1ee3i. C\u00e1c truy v\u1ea5n \u0111\u00e1ng l\u1ebd ph\u1ea3i tr\u1ea3 v\u1ec1 c\u00e1c t\u00e0i li\u1ec7u c\u00f3 li\u00ean quan th\u00ec l\u1ea1i kh\u00f4ng mang l\u1ea1i k\u1ebft qu\u1ea3 ho\u1eb7c n\u1ed9i dung kh\u00f4ng li\u00ean quan. <\/p>\n\n

S\u1ef1 th\u1ea5t b\u1ea1i n\u00e0y th\u1eadt \u0111\u00e1ng th\u1ea5t v\u1ecdng, nh\u01b0ng n\u00f3 c\u0169ng nh\u1ea5n m\u1ea1nh m\u1ed9t b\u00e0i h\u1ecdc quan tr\u1ecdng: ph\u01b0\u01a1ng ph\u00e1p l\u01b0u tr\u1eef c\u0169ng quan tr\u1ecdng nh\u01b0 chi\u1ebfn l\u01b0\u1ee3c truy xu\u1ea5t. T\u00f4i b\u1eaft \u0111\u1ea7u c\u00e2n nh\u1eafc c\u00e1c l\u1ef1a ch\u1ecdn kh\u00e1c, ch\u1eb3ng h\u1ea1n nh\u01b0 s\u1eed d\u1ee5ng MongoDB Atlas \u0111\u1ec3 t\u00ecm ki\u1ebfm vect\u01a1, c\u00f3 kh\u1ea3 n\u0103ng cung c\u1ea5p \u0111\u1ed9 ch\u00ednh x\u00e1c v\u00e0 kh\u1ea3 n\u0103ng m\u1edf r\u1ed9ng m\u00e0 t\u00f4i c\u1ea7n. Tuy nhi\u00ean, tr\u01b0\u1edbc khi cam k\u1ebft v\u1edbi gi\u1ea3i ph\u00e1p thay th\u1ebf n\u00e0y, t\u00f4i mu\u1ed1n kh\u00e1m ph\u00e1 c\u00e1c c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c \u0111\u1ec3 x\u00e1c \u0111\u1ecbnh xem c\u00f3 gi\u1ea3i ph\u00e1p hi\u1ec7u qu\u1ea3 h\u01a1n kh\u00f4ng. <\/p>\n\n

Metadata Retriever v\u00e0 Vector Store: T\u00ecm ki\u1ebfm t\u00ednh \u0111\u1eb7c hi\u1ec7u<\/strong><\/p>\n\n

M\u1ed9t trong nh\u1eefng h\u01b0\u1edbng ti\u1ebfp theo m\u00e0 t\u00f4i kh\u00e1m ph\u00e1 l\u00e0 s\u1eed d\u1ee5ng tr\u00ecnh thu th\u1eadp si\u00eau d\u1eef li\u1ec7u k\u1ebft h\u1ee3p v\u1edbi kho l\u01b0u tr\u1eef vector. \u00dd t\u01b0\u1edfng \u0111\u1eb1ng sau c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y l\u00e0 ph\u00e2n lo\u1ea1i m\u1ea3ng th\u00f4ng tin r\u1ed9ng l\u1edbn trong Fabrice AI v\u00e0 sau \u0111\u00f3 thu th\u1eadp c\u00e2u tr\u1ea3 l\u1eddi d\u1ef1a tr\u00ean c\u00e1c danh m\u1ee5c n\u00e0y. B\u1eb1ng c\u00e1ch c\u1ea5u tr\u00fac d\u1eef li\u1ec7u b\u1eb1ng si\u00eau d\u1eef li\u1ec7u, t\u00f4i hy v\u1ecdng c\u1ea3i thi\u1ec7n kh\u1ea3 n\u0103ng cung c\u1ea5p c\u00e2u tr\u1ea3 l\u1eddi c\u1ee5 th\u1ec3, c\u00f3 m\u1ee5c ti\u00eau c\u1ee7a AI. <\/p>\n\n

Tuy nhi\u00ean, ph\u01b0\u01a1ng ph\u00e1p n\u00e0y c\u0169ng c\u00f3 nh\u1eefng h\u1ea1n ch\u1ebf. M\u1eb7c d\u00f9 c\u00f3 v\u1ebb h\u1ee9a h\u1eb9n tr\u00ean b\u1ec1 m\u1eb7t, AI \u0111\u00e3 ph\u1ea3i v\u1eadt l\u1ed9n \u0111\u1ec3 \u0111\u01b0a ra ph\u1ea3n h\u1ed3i ch\u00ednh x\u00e1c cho m\u1ecdi lo\u1ea1i truy v\u1ea5n. V\u00ed d\u1ee5, khi t\u00f4i h\u1ecfi, “T\u00e1c gi\u1ea3 c\u00f3 l\u1ea1c quan kh\u00f4ng?” H\u1ec7 th\u1ed1ng \u0111\u00e3 kh\u00f4ng di\u1ec5n gi\u1ea3i \u0111\u01b0\u1ee3c c\u00e2u h\u1ecfi trong b\u1ed1i c\u1ea3nh c\u1ee7a n\u1ed9i dung c\u00f3 li\u00ean quan. Thay v\u00ec cung c\u1ea5p ph\u00e2n t\u00edch s\u00e2u s\u1eafc d\u1ef1a tr\u00ean si\u00eau d\u1eef li\u1ec7u, n\u00f3 tr\u1ea3 v\u1ec1 c\u00e2u tr\u1ea3 l\u1eddi m\u01a1 h\u1ed3 ho\u1eb7c kh\u00f4ng c\u00f3 c\u00e2u tr\u1ea3 l\u1eddi n\u00e0o. <\/p>\n\n

C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u00e3 d\u1ea1y cho t\u00f4i m\u1ed9t b\u00e0i h\u1ecdc gi\u00e1 tr\u1ecb v\u1ec1 t\u1ea7m quan tr\u1ecdng c\u1ee7a ng\u1eef c\u1ea3nh trong AI. Ch\u1ec9 ph\u00e2n lo\u1ea1i th\u00f4ng tin th\u00f4i l\u00e0 ch\u01b0a \u0111\u1ee7; AI c\u0169ng ph\u1ea3i hi\u1ec3u c\u00e1ch c\u00e1c danh m\u1ee5c n\u00e0y t\u01b0\u01a1ng t\u00e1c v\u00e0 ch\u1ed3ng ch\u00e9o nhau \u0111\u1ec3 h\u00ecnh th\u00e0nh n\u00ean s\u1ef1 hi\u1ec3u bi\u1ebft g\u1eafn k\u1ebft v\u1ec1 n\u1ed9i dung. N\u1ebfu kh\u00f4ng c\u00f3 chi\u1ec1u s\u00e2u hi\u1ec3u bi\u1ebft n\u00e0y, ngay c\u1ea3 c\u00e1c ph\u01b0\u01a1ng ph\u00e1p truy xu\u1ea5t tinh vi nh\u1ea5t c\u0169ng c\u00f3 th\u1ec3 kh\u00f4ng \u0111\u1ea1t y\u00eau c\u1ea7u. <\/p>\n\n

C\u1ea5u tr\u00fac ki\u1ebfn th\u1ee9c: SummaryTreeIndex<\/strong><\/p>\n\n

Khi t\u00f4i ti\u1ebfp t\u1ee5c tinh ch\u1ec9nh Fabrice AI, t\u00f4i \u0111\u00e3 th\u1eed nghi\u1ec7m t\u1ea1o SummaryTreeIndex. Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y nh\u1eb1m m\u1ee5c \u0111\u00edch t\u00f3m t\u1eaft t\u1ea5t c\u1ea3 c\u00e1c t\u00e0i li\u1ec7u th\u00e0nh \u0111\u1ecbnh d\u1ea1ng c\u00e2y, cho ph\u00e9p AI \u0111i\u1ec1u h\u01b0\u1edbng qua c\u00e1c b\u1ea3n t\u00f3m t\u1eaft n\u00e0y v\u00e0 truy xu\u1ea5t th\u00f4ng tin c\u00f3 li\u00ean quan d\u1ef1a tr\u00ean c\u1ea5u tr\u00fac c\u1ee7a n\u1ed9i dung. <\/p>\n\n

\u00dd t\u01b0\u1edfng l\u00e0 b\u1eb1ng c\u00e1ch t\u00f3m t\u1eaft c\u00e1c t\u00e0i li\u1ec7u, AI c\u00f3 th\u1ec3 nhanh ch\u00f3ng x\u00e1c \u0111\u1ecbnh c\u00e1c \u0111i\u1ec3m ch\u00ednh v\u00e0 tr\u1ea3 l\u1eddi c\u00e1c truy v\u1ea5n b\u1eb1ng th\u00f4ng tin ch\u00ednh x\u00e1c, ng\u1eafn g\u1ecdn. Tuy nhi\u00ean, ph\u01b0\u01a1ng ph\u00e1p n\u00e0y c\u0169ng g\u1eb7p ph\u1ea3i nh\u1eefng th\u00e1ch th\u1ee9c \u0111\u00e1ng k\u1ec3. AI g\u1eb7p kh\u00f3 kh\u0103n trong vi\u1ec7c cung c\u1ea5p c\u00e1c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 \u00fd ngh\u0129a cho c\u00e1c truy v\u1ea5n ph\u1ee9c t\u1ea1p, ch\u1eb3ng h\u1ea1n nh\u01b0 “L\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 \u0111\u01b0a ra quy\u1ebft \u0111\u1ecbnh quan tr\u1ecdng trong cu\u1ed9c s\u1ed1ng?” Thay v\u00ec r\u00fat ra t\u1eeb n\u1ed9i dung phong ph\u00fa, s\u1eafc th\u00e1i \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong c\u00e1c b\u1ea3n t\u00f3m t\u1eaft, ph\u1ea3n h\u1ed3i c\u1ee7a AI th\u01b0\u1eddng h\u1eddi h\u1ee3t ho\u1eb7c kh\u00f4ng \u0111\u1ea7y \u0111\u1ee7. <\/p>\n\n

Kinh nghi\u1ec7m n\u00e0y nh\u1ea5n m\u1ea1nh \u0111\u1ebfn kh\u00f3 kh\u0103n trong vi\u1ec7c c\u00e2n b\u1eb1ng gi\u1eefa chi\u1ec1u r\u1ed9ng v\u00e0 chi\u1ec1u s\u00e2u trong AI. M\u1eb7c d\u00f9 t\u00f3m t\u1eaft c\u00f3 th\u1ec3 cung c\u1ea5p t\u1ed5ng quan c\u1ea5p cao, nh\u01b0ng ch\u00fang th\u01b0\u1eddng thi\u1ebfu b\u1ed1i c\u1ea3nh chi ti\u1ebft c\u1ea7n thi\u1ebft \u0111\u1ec3 tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p h\u01a1n. T\u00f4i nh\u1eadn ra r\u1eb1ng b\u1ea5t k\u1ef3 gi\u1ea3i ph\u00e1p hi\u1ec7u qu\u1ea3 n\u00e0o c\u0169ng c\u1ea7n t\u00edch h\u1ee3p c\u1ea3 n\u1ed9i dung chi ti\u1ebft v\u00e0 t\u00f3m t\u1eaft c\u1ea5p cao, cho ph\u00e9p AI s\u1eed d\u1ee5ng c\u1ea3 hai khi c\u1ea7n. <\/p>\n\n

\u0110\u00e2y l\u00e0 l\u00fd do t\u1ea1i sao trong phi\u00ean b\u1ea3n Fabrice AI hi\u1ec7n t\u1ea1i, tr\u01b0\u1edbc ti\u00ean t\u00f4i \u0111\u1ec3 AI \u0111\u01b0a ra b\u1ea3n t\u00f3m t\u1eaft c\u00e2u tr\u1ea3 l\u1eddi tr\u01b0\u1edbc khi \u0111i s\u00e2u v\u00e0o chi ti\u1ebft.<\/p>\n\n

M\u1edf r\u1ed9ng t\u1ea7m nh\u00ecn: Ch\u1ec9 m\u1ee5c bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c<\/strong><\/p>\n\n

Nh\u1eadn ra nh\u1eefng h\u1ea1n ch\u1ebf c\u1ee7a c\u00e1c ph\u01b0\u01a1ng ph\u00e1p tr\u01b0\u1edbc \u0111\u00e2y, t\u00f4i \u0111\u00e3 chuy\u1ec3n sang m\u1ed9t c\u00e1ch ti\u1ebfp c\u1eadn tinh vi h\u01a1n: Knowledge Graph Index. C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y bao g\u1ed3m vi\u1ec7c x\u00e2y d\u1ef1ng m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c t\u1eeb v\u0103n b\u1ea3n phi c\u1ea5u tr\u00fac, cho ph\u00e9p AI tham gia v\u00e0o truy v\u1ea5n d\u1ef1a tr\u00ean th\u1ef1c th\u1ec3. M\u1ee5c ti\u00eau l\u00e0 t\u1ea1o ra s\u1ef1 hi\u1ec3u bi\u1ebft n\u0103ng \u0111\u1ed9ng v\u00e0 k\u1ebft n\u1ed1i h\u01a1n v\u1ec1 n\u1ed9i dung, cho ph\u00e9p Fabrice AI tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p, n\u1eb7ng v\u1ec1 ng\u1eef c\u1ea3nh hi\u1ec7u qu\u1ea3 h\u01a1n. <\/p>\n\n

M\u1eb7c d\u00f9 c\u00f3 tri\u1ec3n v\u1ecdng, Knowledge Graph Index c\u0169ng g\u1eb7p ph\u1ea3i nh\u1eefng r\u00e0o c\u1ea3n \u0111\u00e1ng k\u1ec3. AI \u0111\u00e3 ph\u1ea3i v\u1eadt l\u1ed9n \u0111\u1ec3 \u0111\u01b0a ra k\u1ebft qu\u1ea3 ch\u00ednh x\u00e1c, \u0111\u1eb7c bi\u1ec7t l\u00e0 \u0111\u1ed1i v\u1edbi c\u00e1c truy v\u1ea5n \u0111\u00f2i h\u1ecfi ph\u1ea3i hi\u1ec3u s\u00e2u v\u1ec1 b\u1ed1i c\u1ea3nh. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “\u0110\u1ecbnh gi\u00e1 Seed & Series A c\u00f4ng b\u1eb1ng l\u00e0 g\u00ec?”, AI m\u1ed9t l\u1ea7n n\u1eefa kh\u00f4ng \u0111\u01b0a ra \u0111\u01b0\u1ee3c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 li\u00ean quan, l\u00e0m n\u1ed5i b\u1eadt s\u1ef1 kh\u00f3 kh\u0103n trong vi\u1ec7c t\u00edch h\u1ee3p v\u0103n b\u1ea3n phi c\u1ea5u tr\u00fac v\u00e0o m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c m\u1ea1ch l\u1ea1c. <\/p>\n\n

C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y, m\u1eb7c d\u00f9 cu\u1ed1i c\u00f9ng kh\u00f4ng th\u00e0nh c\u00f4ng, \u0111\u00e3 cung c\u1ea5p nh\u1eefng hi\u1ec3u bi\u1ebft quan tr\u1ecdng v\u1ec1 nh\u1eefng th\u00e1ch th\u1ee9c khi s\u1eed d\u1ee5ng bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c trong AI. T\u00ednh ph\u1ee9c t\u1ea1p c\u1ee7a d\u1eef li\u1ec7u v\u00e0 nhu c\u1ea7u v\u1ec1 ng\u1eef c\u1ea3nh ch\u00ednh x\u00e1c c\u00f3 ngh\u0129a l\u00e0 ngay c\u1ea3 m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c \u0111\u01b0\u1ee3c x\u00e2y d\u1ef1ng t\u1ed1t c\u0169ng c\u00f3 th\u1ec3 g\u1eb7p kh\u00f3 kh\u0103n trong vi\u1ec7c cung c\u1ea5p k\u1ebft qu\u1ea3 mong mu\u1ed1n. M\u1ed9t nh\u01b0\u1ee3c \u0111i\u1ec3m n\u1eefa c\u1ee7a Knowledge Graph Index l\u00e0 t\u1ed1c \u0111\u1ed9 ch\u1eadm. Th\u1eddi gian ph\u1ea3n h\u1ed3i \u0111\u1ec3 c\u00f3 \u0111\u01b0\u1ee3c c\u00e1c t\u00e0i li\u1ec7u li\u00ean quan r\u1ea5t cao so v\u1edbi ch\u1ec9 m\u1ee5c l\u01b0u tr\u1eef vector. <\/p>\n\n

\u0110\u00e1nh gi\u00e1 l\u1ea1i d\u1eef li\u1ec7u: Gemini<\/strong><\/p>\n\n

Sau nhi\u1ec1u l\u1ea7n th\u1ea5t b\u1ea1i, t\u00f4i quy\u1ebft \u0111\u1ecbnh \u00e1p d\u1ee5ng m\u1ed9t c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c b\u1eb1ng c\u00e1ch t\u1eadn d\u1ee5ng AI c\u1ee7a Google, Gemini<\/a> . \u00dd t\u01b0\u1edfng l\u00e0 t\u1ea1o c\u00e1c t\u1eadp d\u1eef li\u1ec7u t\u1eeb c\u00e1c t\u1ec7p JSON-CSV r\u1ed3i \u0111\u00e0o t\u1ea1o m\u1ed9t m\u00f4 h\u00ecnh LLM t\u00f9y ch\u1ec9nh b\u1eb1ng d\u1eef li\u1ec7u n\u00e0y. T\u00f4i hy v\u1ecdng r\u1eb1ng b\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng d\u1eef li\u1ec7u c\u00f3 c\u1ea5u tr\u00fac v\u00e0 m\u1ed9t m\u00f4 h\u00ecnh \u0111\u00e0o t\u1ea1o m\u1ea1nh m\u1ebd, t\u00f4i c\u00f3 th\u1ec3 v\u01b0\u1ee3t qua m\u1ed9t s\u1ed1 th\u00e1ch th\u1ee9c \u0111\u00e3 t\u1eebng g\u00e2y kh\u00f3 kh\u0103n cho nh\u1eefng l\u1ea7n th\u1eed tr\u01b0\u1edbc. <\/p>\n\n

Tuy nhi\u00ean, c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y c\u0169ng g\u1eb7p ph\u1ea3i kh\u00f3 kh\u0103n. Qu\u00e1 tr\u00ecnh \u0111\u00e0o t\u1ea1o b\u1ecb d\u1eebng l\u1ea1i do \u0111\u1ecbnh d\u1ea1ng d\u1eef li\u1ec7u kh\u00f4ng ch\u00ednh x\u00e1c, khi\u1ebfn m\u00f4 h\u00ecnh kh\u00f4ng \u0111\u01b0\u1ee3c \u0111\u00e0o t\u1ea1o hi\u1ec7u qu\u1ea3. S\u1ef1 c\u1ed1 n\u00e0y nh\u1ea5n m\u1ea1nh t\u1ea7m quan tr\u1ecdng c\u1ee7a t\u00ednh to\u00e0n v\u1eb9n d\u1eef li\u1ec7u trong \u0111\u00e0o t\u1ea1o AI. N\u1ebfu kh\u00f4ng c\u00f3 d\u1eef li\u1ec7u \u0111\u01b0\u1ee3c \u0111\u1ecbnh d\u1ea1ng v\u00e0 c\u1ea5u tr\u00fac \u0111\u00fang c\u00e1ch, ngay c\u1ea3 nh\u1eefng m\u00f4 h\u00ecnh ti\u00ean ti\u1ebfn nh\u1ea5t c\u0169ng kh\u00f4ng th\u1ec3 ho\u1ea1t \u0111\u1ed9ng nh\u01b0 mong \u0111\u1ee3i. <\/p>\n\n

Kinh nghi\u1ec7m n\u00e0y khi\u1ebfn t\u00f4i c\u00e2n nh\u1eafc \u0111\u1ebfn ti\u1ec1m n\u0103ng s\u1eed d\u1ee5ng BigQuery \u0111\u1ec3 l\u01b0u tr\u1eef d\u1eef li\u1ec7u JSON, cung c\u1ea5p n\u1ec1n t\u1ea3ng \u0111\u00e1ng tin c\u1eady v\u00e0 c\u00f3 kh\u1ea3 n\u0103ng m\u1edf r\u1ed9ng h\u01a1n \u0111\u1ec3 qu\u1ea3n l\u00fd c\u00e1c t\u1eadp d\u1eef li\u1ec7u l\u1edbn c\u1ea7n thi\u1ebft \u0111\u1ec3 \u0111\u00e0o t\u1ea1o Fabrice AI m\u1ed9t c\u00e1ch hi\u1ec7u qu\u1ea3.<\/p>\n\n

K\u1ebft h\u1ee3p s\u1ee9c m\u1ea1nh: T\u00e0i li\u1ec7u Langchain v\u1edbi Pinecone<\/strong><\/p>\n\n

B\u1ea5t ch\u1ea5p nh\u1eefng th\u00e1ch th\u1ee9c ph\u1ea3i \u0111\u1ed1i m\u1eb7t cho \u0111\u1ebfn nay, t\u00f4i quy\u1ebft t\u00e2m t\u00ecm ra gi\u1ea3i ph\u00e1p cho ph\u00e9p Fabrice AI l\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t ki\u1ebfn th\u1ee9c hi\u1ec7u qu\u1ea3. Quy\u1ebft t\u00e2m n\u00e0y \u0111\u00e3 d\u1eabn t\u00f4i \u0111\u1ebfn th\u1eed nghi\u1ec7m v\u1edbi c\u00e1c t\u00e0i li\u1ec7u Langchain v\u00e0 Pinecone. Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y bao g\u1ed3m vi\u1ec7c t\u1ea1o m\u1ed9t kho l\u01b0u tr\u1eef vect\u01a1 Pinecone b\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng c\u00e1c t\u00e0i li\u1ec7u Langchain v\u00e0 nh\u00fang OpenAI, sau \u0111\u00f3 truy xu\u1ea5t c\u00e1c t\u00e0i li\u1ec7u t\u01b0\u01a1ng t\u1ef1 h\u00e0ng \u0111\u1ea7u d\u1ef1a tr\u00ean truy v\u1ea5n. <\/p>\n\n

Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y t\u1ecf ra c\u00f3 tri\u1ec3n v\u1ecdng, \u0111\u1eb7c bi\u1ec7t khi truy v\u1ea5n bao g\u1ed3m ti\u00eau \u0111\u1ec1 c\u1ee7a t\u00e0i li\u1ec7u. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “Ch\u00eca kh\u00f3a c\u1ee7a h\u1ea1nh ph\u00fac l\u00e0 g\u00ec?”, AI c\u00f3 th\u1ec3 truy xu\u1ea5t v\u00e0 t\u00f3m t\u1eaft ch\u00ednh x\u00e1c n\u1ed9i dung c\u00f3 li\u00ean quan. Tuy nhi\u00ean, v\u1eabn c\u00f2n nh\u1eefng h\u1ea1n ch\u1ebf, \u0111\u1eb7c bi\u1ec7t khi truy v\u1ea5n thi\u1ebfu c\u00e1c t\u1eeb kh\u00f3a ho\u1eb7c ti\u00eau \u0111\u1ec1 c\u1ee5 th\u1ec3. <\/p>\n\n

C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y ch\u1ee9ng minh ti\u1ec1m n\u0103ng k\u1ebft h\u1ee3p c\u00e1c c\u00f4ng ngh\u1ec7 kh\u00e1c nhau \u0111\u1ec3 n\u00e2ng cao hi\u1ec7u su\u1ea5t AI. B\u1eb1ng c\u00e1ch t\u00edch h\u1ee3p c\u00e1c t\u00e0i li\u1ec7u Langchain v\u1edbi kho l\u01b0u tr\u1eef vector c\u1ee7a Pinecone, t\u00f4i \u0111\u00e3 c\u00f3 th\u1ec3 c\u1ea3i thi\u1ec7n t\u00ednh li\u00ean quan v\u00e0 \u0111\u1ed9 ch\u00ednh x\u00e1c c\u1ee7a ph\u1ea3n h\u1ed3i c\u1ee7a AI, m\u1eb7c d\u00f9 c\u00f3 m\u1ed9t s\u1ed1 h\u1ea1n ch\u1ebf. <\/p>\n\n

\u0110\u1ea1t \u0111\u01b0\u1ee3c s\u1ef1 nh\u1ea5t qu\u00e1n: GPT Builder OpenAI<\/strong><\/p>\n\n

Sau khi kh\u00e1m ph\u00e1 nhi\u1ec1u ph\u01b0\u01a1ng ph\u00e1p v\u00e0 c\u00f4ng ngh\u1ec7 kh\u00e1c nhau, t\u00f4i \u0111\u00e3 chuy\u1ec3n sang GPT Builder<\/a> c\u1ee7a Open AI \u0111\u1ec3 h\u1ee3p nh\u1ea5t v\u00e0 tinh ch\u1ec9nh ki\u1ebfn th\u1ee9c \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong Fabrice AI. B\u1eb1ng c\u00e1ch t\u1ea3i t\u1ea5t c\u1ea3 n\u1ed9i dung v\u00e0o c\u01a1 s\u1edf ki\u1ebfn th\u1ee9c GPT, t\u00f4i h\u01b0\u1edbng \u0111\u1ebfn m\u1ee5c ti\u00eau t\u1ea1o ra m\u1ed9t n\u1ec1n t\u1ea3ng nh\u1ea5t qu\u00e1n v\u00e0 \u0111\u00e1ng tin c\u1eady h\u01a1n \u0111\u1ec3 truy xu\u1ea5t v\u00e0 t\u01b0\u01a1ng t\u00e1c v\u1edbi ki\u1ebfn th\u1ee9c c\u1ee7a m\u00ecnh. <\/p>\n\n

C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u00e3 ch\u1ee9ng minh l\u00e0 m\u1ed9t trong nh\u1eefng c\u00e1ch th\u00e0nh c\u00f4ng nh\u1ea5t, v\u1edbi AI c\u00f3 th\u1ec3 cung c\u1ea5p k\u1ebft qu\u1ea3 t\u1ed1t h\u01a1n tr\u00ean nhi\u1ec1u truy v\u1ea5n. Ch\u00eca kh\u00f3a th\u00e0nh c\u00f4ng n\u00e0y l\u00e0 t\u00edch h\u1ee3p t\u1ea5t c\u1ea3 ki\u1ebfn th\u1ee9c v\u00e0o m\u1ed9t h\u1ec7 th\u1ed1ng th\u1ed1ng nh\u1ea5t, cho ph\u00e9p AI khai th\u00e1c to\u00e0n b\u1ed9 n\u1ed9i dung khi tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi. <\/p>\n\n

Nh\u01b0 \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc, t\u00f4i kh\u00f4ng th\u1ec3 ch\u1ea1y n\u00f3 tr\u00ean trang web c\u1ee7a m\u00ecnh v\u00e0 n\u00f3 ch\u1ec9 kh\u1ea3 d\u1ee5ng v\u1edbi nh\u1eefng ng\u01b0\u1eddi \u0111\u0103ng k\u00fd tr\u1ea3 ph\u00ed c\u1ee7a Chat GPT m\u00e0 t\u00f4i c\u1ea3m th\u1ea5y qu\u00e1 h\u1ea1n ch\u1ebf. Ngo\u00e0i ra, m\u1eb7c d\u00f9 n\u00f3 t\u1ed1t h\u01a1n, t\u00f4i v\u1eabn kh\u00f4ng th\u00edch ch\u1ea5t l\u01b0\u1ee3ng c\u00e2u tr\u1ea3 l\u1eddi v\u00e0 kh\u00f4ng tho\u1ea3i m\u00e1i khi ph\u00e1t h\u00e0nh n\u00f3 cho c\u00f4ng ch\u00fang. <\/p>\n\n

Tinh ch\u1ec9nh cu\u1ed1i c\u00f9ng: Tr\u1ee3 l\u00fd GPT s\u1eed d\u1ee5ng Model 4o<\/strong><\/p>\n\n

M\u1ea3nh gh\u00e9p cu\u1ed1i c\u00f9ng c\u1ee7a c\u00e2u \u0111\u1ed1 trong qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n Fabrice AI xu\u1ea5t hi\u1ec7n c\u00f9ng v\u1edbi s\u1ef1 ra \u0111\u1eddi c\u1ee7a GPT Assistants<\/a> s\u1eed d\u1ee5ng Model 4o. C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u1ea1i di\u1ec7n cho \u0111\u1ec9nh cao c\u1ee7a m\u1ecdi th\u1ee9 t\u00f4i \u0111\u00e3 h\u1ecdc \u0111\u01b0\u1ee3c trong su\u1ed1t d\u1ef1 \u00e1n. B\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng c\u01a1 s\u1edf d\u1eef li\u1ec7u vector v\u00e0 tinh ch\u1ec9nh c\u00e1c l\u1eddi nh\u1eafc, t\u00f4i \u0111\u1eb7t m\u1ee5c ti\u00eau \u0111\u1ea1t \u0111\u01b0\u1ee3c m\u1ee9c \u0111\u1ed9 ch\u00ednh x\u00e1c v\u00e0 hi\u1ec3u bi\u1ebft ng\u1eef c\u1ea3nh cao nh\u1ea5t c\u00f3 th\u1ec3 trong c\u00e1c ph\u1ea3n h\u1ed3i c\u1ee7a AI. <\/p>\n\n

Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y bao g\u1ed3m vi\u1ec7c t\u1ea3i l\u00ean t\u1ea5t c\u1ea3 ki\u1ebfn th\u1ee9c t\u00f4i \u0111\u00e3 t\u00edch l\u0169y \u0111\u01b0\u1ee3c v\u00e0o c\u01a1 s\u1edf d\u1eef li\u1ec7u vector, sau \u0111\u00f3 \u0111\u01b0\u1ee3c s\u1eed d\u1ee5ng l\u00e0m n\u1ec1n t\u1ea3ng cho c\u00e1c t\u01b0\u01a1ng t\u00e1c c\u1ee7a AI. C\u01a1 s\u1edf d\u1eef li\u1ec7u vector cho ph\u00e9p AI th\u1ef1c hi\u1ec7n c\u00e1c t\u00ecm ki\u1ebfm ph\u1ee9c t\u1ea1p h\u01a1n, truy xu\u1ea5t th\u00f4ng tin d\u1ef1a tr\u00ean \u00fd ngh\u0129a ng\u1eef ngh\u0129a c\u1ee7a c\u00e1c truy v\u1ea5n thay v\u00ec ch\u1ec9 d\u1ef1a v\u00e0o vi\u1ec7c kh\u1edbp t\u1eeb kh\u00f3a. \u0110i\u1ec1u n\u00e0y \u0111\u00e1nh d\u1ea5u m\u1ed9t b\u01b0\u1edbc ti\u1ebfn \u0111\u00e1ng k\u1ec3 so v\u1edbi c\u00e1c ph\u01b0\u01a1ng ph\u00e1p ti\u1ebfp c\u1eadn tr\u01b0\u1edbc \u0111\u00e2y, cho ph\u00e9p AI hi\u1ec3u v\u00e0 ph\u1ea3n h\u1ed3i t\u1ed1t h\u01a1n c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p, nhi\u1ec1u s\u1eafc th\u00e1i. <\/p>\n\n

M\u1ed9t trong nh\u1eefng c\u1ea3i ti\u1ebfn ch\u00ednh c\u1ee7a ph\u01b0\u01a1ng ph\u00e1p n\u00e0y l\u00e0 s\u1ef1 tinh ch\u1ec9nh c\u1ea9n th\u1eadn c\u00e1c l\u1eddi nh\u1eafc. B\u1eb1ng c\u00e1ch t\u1ec9 m\u1ec9 t\u1ea1o ra v\u00e0 th\u1eed nghi\u1ec7m c\u00e1c l\u1eddi nh\u1eafc kh\u00e1c nhau, t\u00f4i c\u00f3 th\u1ec3 h\u01b0\u1edbng d\u1eabn AI cung c\u1ea5p c\u00e1c c\u00e2u tr\u1ea3 l\u1eddi ch\u00ednh x\u00e1c v\u00e0 ph\u00f9 h\u1ee3p h\u01a1n. \u0110i\u1ec1u n\u00e0y kh\u00f4ng ch\u1ec9 bao g\u1ed3m vi\u1ec7c \u0111i\u1ec1u ch\u1ec9nh c\u00e1ch di\u1ec5n \u0111\u1ea1t c\u1ee7a c\u00e1c l\u1eddi nh\u1eafc m\u00e0 c\u00f2n th\u1eed nghi\u1ec7m c\u00e1c c\u00e1ch kh\u00e1c nhau \u0111\u1ec3 c\u1ea5u tr\u00fac c\u00e1c truy v\u1ea5n nh\u1eb1m \u0111\u01b0a ra c\u00e1c ph\u1ea3n h\u1ed3i t\u1ed1t nh\u1ea5t c\u00f3 th\u1ec3. K\u1ebft qu\u1ea3 th\u1eadt \u1ea5n t\u01b0\u1ee3ng. AI hi\u1ec7n c\u00f3 th\u1ec3 x\u1eed l\u00fd nhi\u1ec1u lo\u1ea1i truy v\u1ea5n v\u1edbi \u0111\u1ed9 ch\u00ednh x\u00e1c cao, ngay c\u1ea3 khi c\u00e1c c\u00e2u h\u1ecfi m\u1edf ho\u1eb7c \u0111\u00f2i h\u1ecfi hi\u1ec3u bi\u1ebft s\u00e2u s\u1eafc v\u1ec1 b\u1ed1i c\u1ea3nh. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “L\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 \u0111\u01b0a ra nh\u1eefng quy\u1ebft \u0111\u1ecbnh quan tr\u1ecdng nh\u1ea5t trong cu\u1ed9c s\u1ed1ng c\u1ee7a b\u1ea1n?”, AI \u0111\u00e3 \u0111\u01b0a ra c\u00e2u tr\u1ea3 l\u1eddi to\u00e0n di\u1ec7n v\u00e0 s\u00e2u s\u1eafc, d\u1ef1a tr\u00ean nhi\u1ec1u ngu\u1ed3n v\u00e0 g\u00f3c nh\u00ecn kh\u00e1c nhau \u0111\u1ec3 \u0111\u01b0a ra ph\u1ea3n h\u1ed3i to\u00e0n di\u1ec7n. <\/p>\n\n

Th\u00e0nh c\u00f4ng n\u00e0y l\u00e0 \u0111\u1ec9nh cao c\u1ee7a h\u00e0ng tr\u0103m gi\u1edd l\u00e0m vi\u1ec7c v\u00e0 v\u00f4 s\u1ed1 th\u1eed nghi\u1ec7m. N\u00f3 ch\u1ee9ng minh r\u1eb1ng, v\u1edbi s\u1ef1 k\u1ebft h\u1ee3p \u0111\u00fang \u0111\u1eafn gi\u1eefa c\u00f4ng ngh\u1ec7 v\u00e0 s\u1ef1 tinh ch\u1ec9nh, c\u00f3 th\u1ec3 t\u1ea1o ra m\u1ed9t AI kh\u00f4ng ch\u1ec9 c\u00f3 th\u1ec3 l\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t th\u00f4ng tin hi\u1ec7u qu\u1ea3 m\u00e0 c\u00f2n t\u01b0\u01a1ng t\u00e1c v\u1edbi th\u00f4ng tin theo c\u00e1ch c\u00f3 \u00fd ngh\u0129a. S\u1ef1 ph\u00e1t tri\u1ec3n c\u1ee7a GPT Assistants s\u1eed d\u1ee5ng Model 4o \u0111\u00e1nh d\u1ea5u th\u1eddi \u0111i\u1ec3m m\u00e0 Fabrice AI th\u1ef1c s\u1ef1 ph\u00e1t huy h\u1ebft kh\u1ea3 n\u0103ng c\u1ee7a m\u00ecnh, \u0111\u1ea1t \u0111\u01b0\u1ee3c m\u1ee9c \u0111\u1ed9 tinh vi v\u00e0 ch\u00ednh x\u00e1c m\u00e0 t\u00f4i \u0111\u00e3 h\u00ecnh dung ngay t\u1eeb \u0111\u1ea7u. API GPT Assistants sau \u0111\u00f3 \u0111\u00e3 \u0111\u01b0\u1ee3c t\u00edch h\u1ee3p v\u00e0o blog c\u1ee7a t\u00f4i \u0111\u1ec3 cho ph\u00e9p ng\u01b0\u1eddi d\u00f9ng cu\u1ed1i t\u01b0\u01a1ng t\u00e1c v\u1edbi Fabrice AI theo c\u00e1ch b\u1ea1n th\u1ea5y tr\u00ean blog ngay b\u00e2y gi\u1edd. <\/p>\n\n

Suy ng\u1eabm v\u1ec1 h\u00e0nh tr\u00ecnh<\/strong><\/p>\n\n

Qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n Fabrice AI \u0111\u00e3 l\u00e0m n\u1ed5i b\u1eadt s\u1ef1 ph\u1ee9c t\u1ea1p khi l\u00e0m vi\u1ec7c v\u1edbi AI, \u0111\u1eb7c bi\u1ec7t l\u00e0 khi n\u00f3i \u0111\u1ebfn vi\u1ec7c hi\u1ec3u v\u00e0 ng\u1eef c\u1ea3nh h\u00f3a th\u00f4ng tin. N\u00f3 d\u1ea1y t\u00f4i r\u1eb1ng kh\u00f4ng c\u00f3 l\u1ed1i t\u1eaft n\u00e0o trong qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n AI\u2014m\u1ed7i b\u01b0\u1edbc, m\u1ed7i l\u1ea7n l\u1eb7p l\u1ea1i v\u00e0 m\u1ed7i th\u1eed nghi\u1ec7m \u0111\u1ec1u l\u00e0 m\u1ed9t ph\u1ea7n c\u1ea7n thi\u1ebft c\u1ee7a h\u00e0nh tr\u00ecnh h\u01b0\u1edbng t\u1edbi vi\u1ec7c t\u1ea1o ra th\u1ee9 g\u00ec \u0111\u00f3 th\u1ef1c s\u1ef1 hi\u1ec7u qu\u1ea3. <\/p>\n\n

Nh\u00ecn v\u1ec1 ph\u00eda tr\u01b0\u1edbc, t\u00f4i r\u1ea5t h\u00e0o h\u1ee9ng \u0111\u01b0\u1ee3c ti\u1ebfp t\u1ee5c tinh ch\u1ec9nh v\u00e0 m\u1edf r\u1ed9ng Fabrice AI. Nh\u01b0 \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc, t\u00f4i s\u1ebd xem x\u00e9t c\u00e1c c\u00e2u h\u1ecfi \u0111\u01b0\u1ee3c h\u1ecfi \u0111\u1ec3 ho\u00e0n thi\u1ec7n c\u01a1 s\u1edf ki\u1ebfn th\u1ee9c khi c\u00f2n thi\u1ebfu s\u00f3t. T\u00f4i c\u0169ng hy v\u1ecdng cu\u1ed1i c\u00f9ng s\u1ebd ph\u00e1t h\u00e0nh m\u1ed9t phi\u00ean b\u1ea3n t\u01b0\u01a1ng t\u00e1c tr\u00f4ng v\u00e0 nghe gi\u1ed1ng t\u00f4i m\u00e0 b\u1ea1n c\u00f3 th\u1ec3 n\u00f3i chuy\u1ec7n. <\/p>\n","protected":false},"excerpt":{"rendered":"

Nh\u01b0 t\u00f4i \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc , vi\u1ec7c ph\u00e1t tri\u1ec3n Fabrice AI ph\u1ee9c t\u1ea1p h\u01a1n d\u1ef1 ki\u1ebfn r\u1ea5t nhi\u1ec1u, bu\u1ed9c t\u00f4i ph\u1ea3i kh\u00e1m ph\u00e1 nhi\u1ec1u c\u00e1ch ti\u1ebfp … Continue reading “Fabrice AI: H\u00e0nh tr\u00ecnh k\u1ef9 thu\u1eadt”<\/span><\/a><\/p>\n","protected":false},"author":9,"featured_media":46234,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"footnotes":""},"categories":[238,927],"tags":[],"class_list":["post-46237","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-suy-nghi-ca-nhan","category-tien-ich-cong-nghe"],"acf":[],"contentUpdated":"Fabrice AI: H\u00e0nh tr\u00ecnh k\u1ef9 thu\u1eadt. Categories - Suy ngh\u0129 c\u00e1 nh\u00e2n, Ti\u1ec7n \u00edch c\u00f4ng ngh\u1ec7. Date-Posted - 2024-09-04T14:05:16 . \n Nh\u01b0 t\u00f4i \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc , vi\u1ec7c ph\u00e1t tri\u1ec3n Fabrice AI ph\u1ee9c t\u1ea1p h\u01a1n d\u1ef1 ki\u1ebfn r\u1ea5t nhi\u1ec1u, bu\u1ed9c t\u00f4i ph\u1ea3i kh\u00e1m ph\u00e1 nhi\u1ec1u c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c nhau.\n C\u00e1ch ti\u1ebfp c\u1eadn ban \u0111\u1ea7u: Ch\u1ec9 s\u1ed1 Llama – T\u00ecm ki\u1ebfm Vector\n L\u1ea7n \u0111\u1ea7u ti\u00ean t\u00f4i th\u1eed nghi\u1ec7m n\u00e2ng cao kh\u1ea3 n\u0103ng truy xu\u1ea5t c\u1ee7a Fabrice AI li\u00ean quan \u0111\u1ebfn vi\u1ec7c s\u1eed d\u1ee5ng Llama Index \u0111\u1ec3 t\u00ecm ki\u1ebfm vector. Kh\u00e1i ni\u1ec7m n\u00e0y r\u1ea5t \u0111\u01a1n gi\u1ea3n: l\u1ea5y n\u1ed9i dung t\u1eeb blog c\u1ee7a t\u00f4i, chuy\u1ec3n \u0111\u1ed5i th\u00e0nh t\u00e0i li\u1ec7u Langchain, sau \u0111\u00f3 chuy\u1ec3n \u0111\u1ed5i ch\u00fang th\u00e0nh t\u00e0i li\u1ec7u Llama. C\u00e1c t\u00e0i li\u1ec7u Llama n\u00e0y sau \u0111\u00f3 s\u1ebd \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong m\u1ed9t vector index, cho ph\u00e9p t\u00f4i truy v\u1ea5n ch\u1ec9 m\u1ee5c n\u00e0y \u0111\u1ec3 t\u00ecm th\u00f4ng tin c\u00f3 li\u00ean quan. \n Tuy nhi\u00ean, khi t\u00f4i b\u1eaft \u0111\u1ea7u th\u1eed nghi\u1ec7m h\u1ec7 th\u1ed1ng, t\u00f4i nh\u1eadn ra r\u1eb1ng c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y kh\u00f4ng mang l\u1ea1i k\u1ebft qu\u1ea3 nh\u01b0 mong \u0111\u1ee3i. C\u1ee5 th\u1ec3, khi t\u00f4i truy v\u1ea5n h\u1ec7 th\u1ed1ng b\u1eb1ng nh\u1eefng c\u00e2u h\u1ecfi n\u1eb7ng v\u1ec1 ng\u1eef c\u1ea3nh nh\u01b0 “Nh\u1eefng sai l\u1ea7m l\u1edbn nh\u1ea5t m\u00e0 ng\u01b0\u1eddi s\u00e1ng l\u1eadp th\u1ecb tr\u01b0\u1eddng m\u1eafc ph\u1ea3i l\u00e0 g\u00ec?”, AI \u0111\u00e3 kh\u00f4ng \u0111\u01b0a ra \u0111\u01b0\u1ee3c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 \u00fd ngh\u0129a. Thay v\u00ec truy xu\u1ea5t n\u1ed9i dung s\u1eafc th\u00e1i m\u00e0 t\u00f4i bi\u1ebft l\u00e0 \u0111\u00e3 \u0111\u01b0\u1ee3c nh\u00fang trong d\u1eef li\u1ec7u, n\u00f3 tr\u1ea3 v\u1ec1 nh\u1eefng ph\u1ea3n h\u1ed3i kh\u00f4ng li\u00ean quan ho\u1eb7c kh\u00f4ng \u0111\u1ea7y \u0111\u1ee7. \n Th\u1ea5t b\u1ea1i ban \u0111\u1ea7u n\u00e0y khi\u1ebfn t\u00f4i ph\u1ea3i xem x\u00e9t l\u1ea1i c\u00e1ch ti\u1ebfp c\u1eadn c\u1ee7a m\u00ecnh. T\u00f4i nh\u1eadn ra r\u1eb1ng ch\u1ec9 l\u01b0u tr\u1eef n\u1ed9i dung trong m\u1ed9t ch\u1ec9 m\u1ee5c vect\u01a1 l\u00e0 kh\u00f4ng \u0111\u1ee7; c\u01a1 ch\u1ebf truy xu\u1ea5t c\u1ea7n ph\u1ea3i hi\u1ec3u b\u1ed1i c\u1ea3nh v\u00e0 s\u1eafc th\u00e1i c\u1ee7a c\u00e1c c\u00e2u h\u1ecfi \u0111\u01b0\u1ee3c h\u1ecfi. Nh\u1eadn th\u1ee9c n\u00e0y l\u00e0 b\u00e0i h\u1ecdc \u0111\u1ea7u ti\u00ean trong s\u1ed1 nhi\u1ec1u b\u00e0i h\u1ecdc s\u1ebd \u0111\u1ecbnh h\u00ecnh s\u1ef1 ph\u00e1t tri\u1ec3n c\u1ee7a Fabrice AI. \n L\u01b0u tr\u1eef ki\u1ebfn th\u1ee9c: L\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t t\u00e0i li\u1ec7u MongoDB\n V\u1edbi nh\u1eefng h\u1ea1n ch\u1ebf c\u1ee7a ph\u01b0\u01a1ng ph\u00e1p Llama Index, ti\u1ebfp theo t\u00f4i kh\u00e1m ph\u00e1 c\u00e1ch l\u01b0u tr\u1eef c\u00e1c t\u00e0i li\u1ec7u Llama trong MongoDB. S\u01a1 \u0111\u1ed3 linh ho\u1ea1t v\u00e0 c\u1ea5u tr\u00fac h\u01b0\u1edbng t\u00e0i li\u1ec7u c\u1ee7a MongoDB c\u00f3 v\u1ebb l\u00e0 gi\u1ea3i ph\u00e1p \u0111\u1ea7y h\u1ee9a h\u1eb9n \u0111\u1ec3 qu\u1ea3n l\u00fd nhi\u1ec1u lo\u1ea1i n\u1ed9i dung kh\u00e1c nhau m\u00e0 t\u00f4i \u0111\u00e3 t\u00edch l\u0169y trong nhi\u1ec1u n\u0103m. \n K\u1ebf ho\u1ea1ch l\u00e0 t\u1ea1o ra tr\u1ea3i nghi\u1ec7m t\u00ecm ki\u1ebfm n\u0103ng \u0111\u1ed9ng v\u00e0 ph\u1ea3n h\u1ed3i h\u01a1n. Tuy nhi\u00ean, c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y nhanh ch\u00f3ng g\u1eb7p ph\u1ea3i v\u1ea5n \u0111\u1ec1. Ch\u1ee9c n\u0103ng t\u00ecm ki\u1ebfm m\u00e0 t\u00f4i d\u1ef1 \u0111o\u00e1n s\u1ebd m\u1ea1nh m\u1ebd h\u01a1n, \u0111\u00e3 kh\u00f4ng ho\u1ea1t \u0111\u1ed9ng nh\u01b0 mong \u0111\u1ee3i. C\u00e1c truy v\u1ea5n \u0111\u00e1ng l\u1ebd ph\u1ea3i tr\u1ea3 v\u1ec1 c\u00e1c t\u00e0i li\u1ec7u c\u00f3 li\u00ean quan th\u00ec l\u1ea1i kh\u00f4ng mang l\u1ea1i k\u1ebft qu\u1ea3 ho\u1eb7c n\u1ed9i dung kh\u00f4ng li\u00ean quan. \n S\u1ef1 th\u1ea5t b\u1ea1i n\u00e0y th\u1eadt \u0111\u00e1ng th\u1ea5t v\u1ecdng, nh\u01b0ng n\u00f3 c\u0169ng nh\u1ea5n m\u1ea1nh m\u1ed9t b\u00e0i h\u1ecdc quan tr\u1ecdng: ph\u01b0\u01a1ng ph\u00e1p l\u01b0u tr\u1eef c\u0169ng quan tr\u1ecdng nh\u01b0 chi\u1ebfn l\u01b0\u1ee3c truy xu\u1ea5t. T\u00f4i b\u1eaft \u0111\u1ea7u c\u00e2n nh\u1eafc c\u00e1c l\u1ef1a ch\u1ecdn kh\u00e1c, ch\u1eb3ng h\u1ea1n nh\u01b0 s\u1eed d\u1ee5ng MongoDB Atlas \u0111\u1ec3 t\u00ecm ki\u1ebfm vect\u01a1, c\u00f3 kh\u1ea3 n\u0103ng cung c\u1ea5p \u0111\u1ed9 ch\u00ednh x\u00e1c v\u00e0 kh\u1ea3 n\u0103ng m\u1edf r\u1ed9ng m\u00e0 t\u00f4i c\u1ea7n. Tuy nhi\u00ean, tr\u01b0\u1edbc khi cam k\u1ebft v\u1edbi gi\u1ea3i ph\u00e1p thay th\u1ebf n\u00e0y, t\u00f4i mu\u1ed1n kh\u00e1m ph\u00e1 c\u00e1c c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c \u0111\u1ec3 x\u00e1c \u0111\u1ecbnh xem c\u00f3 gi\u1ea3i ph\u00e1p hi\u1ec7u qu\u1ea3 h\u01a1n kh\u00f4ng. \n Metadata Retriever v\u00e0 Vector Store: T\u00ecm ki\u1ebfm t\u00ednh \u0111\u1eb7c hi\u1ec7u\n M\u1ed9t trong nh\u1eefng h\u01b0\u1edbng ti\u1ebfp theo m\u00e0 t\u00f4i kh\u00e1m ph\u00e1 l\u00e0 s\u1eed d\u1ee5ng tr\u00ecnh thu th\u1eadp si\u00eau d\u1eef li\u1ec7u k\u1ebft h\u1ee3p v\u1edbi kho l\u01b0u tr\u1eef vector. \u00dd t\u01b0\u1edfng \u0111\u1eb1ng sau c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y l\u00e0 ph\u00e2n lo\u1ea1i m\u1ea3ng th\u00f4ng tin r\u1ed9ng l\u1edbn trong Fabrice AI v\u00e0 sau \u0111\u00f3 thu th\u1eadp c\u00e2u tr\u1ea3 l\u1eddi d\u1ef1a tr\u00ean c\u00e1c danh m\u1ee5c n\u00e0y. B\u1eb1ng c\u00e1ch c\u1ea5u tr\u00fac d\u1eef li\u1ec7u b\u1eb1ng si\u00eau d\u1eef li\u1ec7u, t\u00f4i hy v\u1ecdng c\u1ea3i thi\u1ec7n kh\u1ea3 n\u0103ng cung c\u1ea5p c\u00e2u tr\u1ea3 l\u1eddi c\u1ee5 th\u1ec3, c\u00f3 m\u1ee5c ti\u00eau c\u1ee7a AI. \n Tuy nhi\u00ean, ph\u01b0\u01a1ng ph\u00e1p n\u00e0y c\u0169ng c\u00f3 nh\u1eefng h\u1ea1n ch\u1ebf. M\u1eb7c d\u00f9 c\u00f3 v\u1ebb h\u1ee9a h\u1eb9n tr\u00ean b\u1ec1 m\u1eb7t, AI \u0111\u00e3 ph\u1ea3i v\u1eadt l\u1ed9n \u0111\u1ec3 \u0111\u01b0a ra ph\u1ea3n h\u1ed3i ch\u00ednh x\u00e1c cho m\u1ecdi lo\u1ea1i truy v\u1ea5n. V\u00ed d\u1ee5, khi t\u00f4i h\u1ecfi, “T\u00e1c gi\u1ea3 c\u00f3 l\u1ea1c quan kh\u00f4ng?” H\u1ec7 th\u1ed1ng \u0111\u00e3 kh\u00f4ng di\u1ec5n gi\u1ea3i \u0111\u01b0\u1ee3c c\u00e2u h\u1ecfi trong b\u1ed1i c\u1ea3nh c\u1ee7a n\u1ed9i dung c\u00f3 li\u00ean quan. Thay v\u00ec cung c\u1ea5p ph\u00e2n t\u00edch s\u00e2u s\u1eafc d\u1ef1a tr\u00ean si\u00eau d\u1eef li\u1ec7u, n\u00f3 tr\u1ea3 v\u1ec1 c\u00e2u tr\u1ea3 l\u1eddi m\u01a1 h\u1ed3 ho\u1eb7c kh\u00f4ng c\u00f3 c\u00e2u tr\u1ea3 l\u1eddi n\u00e0o. \n C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u00e3 d\u1ea1y cho t\u00f4i m\u1ed9t b\u00e0i h\u1ecdc gi\u00e1 tr\u1ecb v\u1ec1 t\u1ea7m quan tr\u1ecdng c\u1ee7a ng\u1eef c\u1ea3nh trong AI. Ch\u1ec9 ph\u00e2n lo\u1ea1i th\u00f4ng tin th\u00f4i l\u00e0 ch\u01b0a \u0111\u1ee7; AI c\u0169ng ph\u1ea3i hi\u1ec3u c\u00e1ch c\u00e1c danh m\u1ee5c n\u00e0y t\u01b0\u01a1ng t\u00e1c v\u00e0 ch\u1ed3ng ch\u00e9o nhau \u0111\u1ec3 h\u00ecnh th\u00e0nh n\u00ean s\u1ef1 hi\u1ec3u bi\u1ebft g\u1eafn k\u1ebft v\u1ec1 n\u1ed9i dung. N\u1ebfu kh\u00f4ng c\u00f3 chi\u1ec1u s\u00e2u hi\u1ec3u bi\u1ebft n\u00e0y, ngay c\u1ea3 c\u00e1c ph\u01b0\u01a1ng ph\u00e1p truy xu\u1ea5t tinh vi nh\u1ea5t c\u0169ng c\u00f3 th\u1ec3 kh\u00f4ng \u0111\u1ea1t y\u00eau c\u1ea7u. \n C\u1ea5u tr\u00fac ki\u1ebfn th\u1ee9c: SummaryTreeIndex\n Khi t\u00f4i ti\u1ebfp t\u1ee5c tinh ch\u1ec9nh Fabrice AI, t\u00f4i \u0111\u00e3 th\u1eed nghi\u1ec7m t\u1ea1o SummaryTreeIndex. Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y nh\u1eb1m m\u1ee5c \u0111\u00edch t\u00f3m t\u1eaft t\u1ea5t c\u1ea3 c\u00e1c t\u00e0i li\u1ec7u th\u00e0nh \u0111\u1ecbnh d\u1ea1ng c\u00e2y, cho ph\u00e9p AI \u0111i\u1ec1u h\u01b0\u1edbng qua c\u00e1c b\u1ea3n t\u00f3m t\u1eaft n\u00e0y v\u00e0 truy xu\u1ea5t th\u00f4ng tin c\u00f3 li\u00ean quan d\u1ef1a tr\u00ean c\u1ea5u tr\u00fac c\u1ee7a n\u1ed9i dung. \n \u00dd t\u01b0\u1edfng l\u00e0 b\u1eb1ng c\u00e1ch t\u00f3m t\u1eaft c\u00e1c t\u00e0i li\u1ec7u, AI c\u00f3 th\u1ec3 nhanh ch\u00f3ng x\u00e1c \u0111\u1ecbnh c\u00e1c \u0111i\u1ec3m ch\u00ednh v\u00e0 tr\u1ea3 l\u1eddi c\u00e1c truy v\u1ea5n b\u1eb1ng th\u00f4ng tin ch\u00ednh x\u00e1c, ng\u1eafn g\u1ecdn. Tuy nhi\u00ean, ph\u01b0\u01a1ng ph\u00e1p n\u00e0y c\u0169ng g\u1eb7p ph\u1ea3i nh\u1eefng th\u00e1ch th\u1ee9c \u0111\u00e1ng k\u1ec3. AI g\u1eb7p kh\u00f3 kh\u0103n trong vi\u1ec7c cung c\u1ea5p c\u00e1c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 \u00fd ngh\u0129a cho c\u00e1c truy v\u1ea5n ph\u1ee9c t\u1ea1p, ch\u1eb3ng h\u1ea1n nh\u01b0 “L\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 \u0111\u01b0a ra quy\u1ebft \u0111\u1ecbnh quan tr\u1ecdng trong cu\u1ed9c s\u1ed1ng?” Thay v\u00ec r\u00fat ra t\u1eeb n\u1ed9i dung phong ph\u00fa, s\u1eafc th\u00e1i \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong c\u00e1c b\u1ea3n t\u00f3m t\u1eaft, ph\u1ea3n h\u1ed3i c\u1ee7a AI th\u01b0\u1eddng h\u1eddi h\u1ee3t ho\u1eb7c kh\u00f4ng \u0111\u1ea7y \u0111\u1ee7. \n Kinh nghi\u1ec7m n\u00e0y nh\u1ea5n m\u1ea1nh \u0111\u1ebfn kh\u00f3 kh\u0103n trong vi\u1ec7c c\u00e2n b\u1eb1ng gi\u1eefa chi\u1ec1u r\u1ed9ng v\u00e0 chi\u1ec1u s\u00e2u trong AI. M\u1eb7c d\u00f9 t\u00f3m t\u1eaft c\u00f3 th\u1ec3 cung c\u1ea5p t\u1ed5ng quan c\u1ea5p cao, nh\u01b0ng ch\u00fang th\u01b0\u1eddng thi\u1ebfu b\u1ed1i c\u1ea3nh chi ti\u1ebft c\u1ea7n thi\u1ebft \u0111\u1ec3 tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p h\u01a1n. T\u00f4i nh\u1eadn ra r\u1eb1ng b\u1ea5t k\u1ef3 gi\u1ea3i ph\u00e1p hi\u1ec7u qu\u1ea3 n\u00e0o c\u0169ng c\u1ea7n t\u00edch h\u1ee3p c\u1ea3 n\u1ed9i dung chi ti\u1ebft v\u00e0 t\u00f3m t\u1eaft c\u1ea5p cao, cho ph\u00e9p AI s\u1eed d\u1ee5ng c\u1ea3 hai khi c\u1ea7n. \n \u0110\u00e2y l\u00e0 l\u00fd do t\u1ea1i sao trong phi\u00ean b\u1ea3n Fabrice AI hi\u1ec7n t\u1ea1i, tr\u01b0\u1edbc ti\u00ean t\u00f4i \u0111\u1ec3 AI \u0111\u01b0a ra b\u1ea3n t\u00f3m t\u1eaft c\u00e2u tr\u1ea3 l\u1eddi tr\u01b0\u1edbc khi \u0111i s\u00e2u v\u00e0o chi ti\u1ebft.\n M\u1edf r\u1ed9ng t\u1ea7m nh\u00ecn: Ch\u1ec9 m\u1ee5c bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c\n Nh\u1eadn ra nh\u1eefng h\u1ea1n ch\u1ebf c\u1ee7a c\u00e1c ph\u01b0\u01a1ng ph\u00e1p tr\u01b0\u1edbc \u0111\u00e2y, t\u00f4i \u0111\u00e3 chuy\u1ec3n sang m\u1ed9t c\u00e1ch ti\u1ebfp c\u1eadn tinh vi h\u01a1n: Knowledge Graph Index. C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y bao g\u1ed3m vi\u1ec7c x\u00e2y d\u1ef1ng m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c t\u1eeb v\u0103n b\u1ea3n phi c\u1ea5u tr\u00fac, cho ph\u00e9p AI tham gia v\u00e0o truy v\u1ea5n d\u1ef1a tr\u00ean th\u1ef1c th\u1ec3. M\u1ee5c ti\u00eau l\u00e0 t\u1ea1o ra s\u1ef1 hi\u1ec3u bi\u1ebft n\u0103ng \u0111\u1ed9ng v\u00e0 k\u1ebft n\u1ed1i h\u01a1n v\u1ec1 n\u1ed9i dung, cho ph\u00e9p Fabrice AI tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p, n\u1eb7ng v\u1ec1 ng\u1eef c\u1ea3nh hi\u1ec7u qu\u1ea3 h\u01a1n. \n M\u1eb7c d\u00f9 c\u00f3 tri\u1ec3n v\u1ecdng, Knowledge Graph Index c\u0169ng g\u1eb7p ph\u1ea3i nh\u1eefng r\u00e0o c\u1ea3n \u0111\u00e1ng k\u1ec3. AI \u0111\u00e3 ph\u1ea3i v\u1eadt l\u1ed9n \u0111\u1ec3 \u0111\u01b0a ra k\u1ebft qu\u1ea3 ch\u00ednh x\u00e1c, \u0111\u1eb7c bi\u1ec7t l\u00e0 \u0111\u1ed1i v\u1edbi c\u00e1c truy v\u1ea5n \u0111\u00f2i h\u1ecfi ph\u1ea3i hi\u1ec3u s\u00e2u v\u1ec1 b\u1ed1i c\u1ea3nh. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “\u0110\u1ecbnh gi\u00e1 Seed & Series A c\u00f4ng b\u1eb1ng l\u00e0 g\u00ec?”, AI m\u1ed9t l\u1ea7n n\u1eefa kh\u00f4ng \u0111\u01b0a ra \u0111\u01b0\u1ee3c c\u00e2u tr\u1ea3 l\u1eddi c\u00f3 li\u00ean quan, l\u00e0m n\u1ed5i b\u1eadt s\u1ef1 kh\u00f3 kh\u0103n trong vi\u1ec7c t\u00edch h\u1ee3p v\u0103n b\u1ea3n phi c\u1ea5u tr\u00fac v\u00e0o m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c m\u1ea1ch l\u1ea1c. \n C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y, m\u1eb7c d\u00f9 cu\u1ed1i c\u00f9ng kh\u00f4ng th\u00e0nh c\u00f4ng, \u0111\u00e3 cung c\u1ea5p nh\u1eefng hi\u1ec3u bi\u1ebft quan tr\u1ecdng v\u1ec1 nh\u1eefng th\u00e1ch th\u1ee9c khi s\u1eed d\u1ee5ng bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c trong AI. T\u00ednh ph\u1ee9c t\u1ea1p c\u1ee7a d\u1eef li\u1ec7u v\u00e0 nhu c\u1ea7u v\u1ec1 ng\u1eef c\u1ea3nh ch\u00ednh x\u00e1c c\u00f3 ngh\u0129a l\u00e0 ngay c\u1ea3 m\u1ed9t bi\u1ec3u \u0111\u1ed3 ki\u1ebfn th\u1ee9c \u0111\u01b0\u1ee3c x\u00e2y d\u1ef1ng t\u1ed1t c\u0169ng c\u00f3 th\u1ec3 g\u1eb7p kh\u00f3 kh\u0103n trong vi\u1ec7c cung c\u1ea5p k\u1ebft qu\u1ea3 mong mu\u1ed1n. M\u1ed9t nh\u01b0\u1ee3c \u0111i\u1ec3m n\u1eefa c\u1ee7a Knowledge Graph Index l\u00e0 t\u1ed1c \u0111\u1ed9 ch\u1eadm. Th\u1eddi gian ph\u1ea3n h\u1ed3i \u0111\u1ec3 c\u00f3 \u0111\u01b0\u1ee3c c\u00e1c t\u00e0i li\u1ec7u li\u00ean quan r\u1ea5t cao so v\u1edbi ch\u1ec9 m\u1ee5c l\u01b0u tr\u1eef vector. \n \u0110\u00e1nh gi\u00e1 l\u1ea1i d\u1eef li\u1ec7u: Gemini\n Sau nhi\u1ec1u l\u1ea7n th\u1ea5t b\u1ea1i, t\u00f4i quy\u1ebft \u0111\u1ecbnh \u00e1p d\u1ee5ng m\u1ed9t c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c b\u1eb1ng c\u00e1ch t\u1eadn d\u1ee5ng AI c\u1ee7a Google, Gemini . \u00dd t\u01b0\u1edfng l\u00e0 t\u1ea1o c\u00e1c t\u1eadp d\u1eef li\u1ec7u t\u1eeb c\u00e1c t\u1ec7p JSON-CSV r\u1ed3i \u0111\u00e0o t\u1ea1o m\u1ed9t m\u00f4 h\u00ecnh LLM t\u00f9y ch\u1ec9nh b\u1eb1ng d\u1eef li\u1ec7u n\u00e0y. T\u00f4i hy v\u1ecdng r\u1eb1ng b\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng d\u1eef li\u1ec7u c\u00f3 c\u1ea5u tr\u00fac v\u00e0 m\u1ed9t m\u00f4 h\u00ecnh \u0111\u00e0o t\u1ea1o m\u1ea1nh m\u1ebd, t\u00f4i c\u00f3 th\u1ec3 v\u01b0\u1ee3t qua m\u1ed9t s\u1ed1 th\u00e1ch th\u1ee9c \u0111\u00e3 t\u1eebng g\u00e2y kh\u00f3 kh\u0103n cho nh\u1eefng l\u1ea7n th\u1eed tr\u01b0\u1edbc. \n Tuy nhi\u00ean, c\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y c\u0169ng g\u1eb7p ph\u1ea3i kh\u00f3 kh\u0103n. Qu\u00e1 tr\u00ecnh \u0111\u00e0o t\u1ea1o b\u1ecb d\u1eebng l\u1ea1i do \u0111\u1ecbnh d\u1ea1ng d\u1eef li\u1ec7u kh\u00f4ng ch\u00ednh x\u00e1c, khi\u1ebfn m\u00f4 h\u00ecnh kh\u00f4ng \u0111\u01b0\u1ee3c \u0111\u00e0o t\u1ea1o hi\u1ec7u qu\u1ea3. S\u1ef1 c\u1ed1 n\u00e0y nh\u1ea5n m\u1ea1nh t\u1ea7m quan tr\u1ecdng c\u1ee7a t\u00ednh to\u00e0n v\u1eb9n d\u1eef li\u1ec7u trong \u0111\u00e0o t\u1ea1o AI. N\u1ebfu kh\u00f4ng c\u00f3 d\u1eef li\u1ec7u \u0111\u01b0\u1ee3c \u0111\u1ecbnh d\u1ea1ng v\u00e0 c\u1ea5u tr\u00fac \u0111\u00fang c\u00e1ch, ngay c\u1ea3 nh\u1eefng m\u00f4 h\u00ecnh ti\u00ean ti\u1ebfn nh\u1ea5t c\u0169ng kh\u00f4ng th\u1ec3 ho\u1ea1t \u0111\u1ed9ng nh\u01b0 mong \u0111\u1ee3i. \n Kinh nghi\u1ec7m n\u00e0y khi\u1ebfn t\u00f4i c\u00e2n nh\u1eafc \u0111\u1ebfn ti\u1ec1m n\u0103ng s\u1eed d\u1ee5ng BigQuery \u0111\u1ec3 l\u01b0u tr\u1eef d\u1eef li\u1ec7u JSON, cung c\u1ea5p n\u1ec1n t\u1ea3ng \u0111\u00e1ng tin c\u1eady v\u00e0 c\u00f3 kh\u1ea3 n\u0103ng m\u1edf r\u1ed9ng h\u01a1n \u0111\u1ec3 qu\u1ea3n l\u00fd c\u00e1c t\u1eadp d\u1eef li\u1ec7u l\u1edbn c\u1ea7n thi\u1ebft \u0111\u1ec3 \u0111\u00e0o t\u1ea1o Fabrice AI m\u1ed9t c\u00e1ch hi\u1ec7u qu\u1ea3.\n K\u1ebft h\u1ee3p s\u1ee9c m\u1ea1nh: T\u00e0i li\u1ec7u Langchain v\u1edbi Pinecone\n B\u1ea5t ch\u1ea5p nh\u1eefng th\u00e1ch th\u1ee9c ph\u1ea3i \u0111\u1ed1i m\u1eb7t cho \u0111\u1ebfn nay, t\u00f4i quy\u1ebft t\u00e2m t\u00ecm ra gi\u1ea3i ph\u00e1p cho ph\u00e9p Fabrice AI l\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t ki\u1ebfn th\u1ee9c hi\u1ec7u qu\u1ea3. Quy\u1ebft t\u00e2m n\u00e0y \u0111\u00e3 d\u1eabn t\u00f4i \u0111\u1ebfn th\u1eed nghi\u1ec7m v\u1edbi c\u00e1c t\u00e0i li\u1ec7u Langchain v\u00e0 Pinecone. Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y bao g\u1ed3m vi\u1ec7c t\u1ea1o m\u1ed9t kho l\u01b0u tr\u1eef vect\u01a1 Pinecone b\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng c\u00e1c t\u00e0i li\u1ec7u Langchain v\u00e0 nh\u00fang OpenAI, sau \u0111\u00f3 truy xu\u1ea5t c\u00e1c t\u00e0i li\u1ec7u t\u01b0\u01a1ng t\u1ef1 h\u00e0ng \u0111\u1ea7u d\u1ef1a tr\u00ean truy v\u1ea5n. \n Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y t\u1ecf ra c\u00f3 tri\u1ec3n v\u1ecdng, \u0111\u1eb7c bi\u1ec7t khi truy v\u1ea5n bao g\u1ed3m ti\u00eau \u0111\u1ec1 c\u1ee7a t\u00e0i li\u1ec7u. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “Ch\u00eca kh\u00f3a c\u1ee7a h\u1ea1nh ph\u00fac l\u00e0 g\u00ec?”, AI c\u00f3 th\u1ec3 truy xu\u1ea5t v\u00e0 t\u00f3m t\u1eaft ch\u00ednh x\u00e1c n\u1ed9i dung c\u00f3 li\u00ean quan. Tuy nhi\u00ean, v\u1eabn c\u00f2n nh\u1eefng h\u1ea1n ch\u1ebf, \u0111\u1eb7c bi\u1ec7t khi truy v\u1ea5n thi\u1ebfu c\u00e1c t\u1eeb kh\u00f3a ho\u1eb7c ti\u00eau \u0111\u1ec1 c\u1ee5 th\u1ec3. \n C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y ch\u1ee9ng minh ti\u1ec1m n\u0103ng k\u1ebft h\u1ee3p c\u00e1c c\u00f4ng ngh\u1ec7 kh\u00e1c nhau \u0111\u1ec3 n\u00e2ng cao hi\u1ec7u su\u1ea5t AI. B\u1eb1ng c\u00e1ch t\u00edch h\u1ee3p c\u00e1c t\u00e0i li\u1ec7u Langchain v\u1edbi kho l\u01b0u tr\u1eef vector c\u1ee7a Pinecone, t\u00f4i \u0111\u00e3 c\u00f3 th\u1ec3 c\u1ea3i thi\u1ec7n t\u00ednh li\u00ean quan v\u00e0 \u0111\u1ed9 ch\u00ednh x\u00e1c c\u1ee7a ph\u1ea3n h\u1ed3i c\u1ee7a AI, m\u1eb7c d\u00f9 c\u00f3 m\u1ed9t s\u1ed1 h\u1ea1n ch\u1ebf. \n \u0110\u1ea1t \u0111\u01b0\u1ee3c s\u1ef1 nh\u1ea5t qu\u00e1n: GPT Builder OpenAI\n Sau khi kh\u00e1m ph\u00e1 nhi\u1ec1u ph\u01b0\u01a1ng ph\u00e1p v\u00e0 c\u00f4ng ngh\u1ec7 kh\u00e1c nhau, t\u00f4i \u0111\u00e3 chuy\u1ec3n sang GPT Builder c\u1ee7a Open AI \u0111\u1ec3 h\u1ee3p nh\u1ea5t v\u00e0 tinh ch\u1ec9nh ki\u1ebfn th\u1ee9c \u0111\u01b0\u1ee3c l\u01b0u tr\u1eef trong Fabrice AI. B\u1eb1ng c\u00e1ch t\u1ea3i t\u1ea5t c\u1ea3 n\u1ed9i dung v\u00e0o c\u01a1 s\u1edf ki\u1ebfn th\u1ee9c GPT, t\u00f4i h\u01b0\u1edbng \u0111\u1ebfn m\u1ee5c ti\u00eau t\u1ea1o ra m\u1ed9t n\u1ec1n t\u1ea3ng nh\u1ea5t qu\u00e1n v\u00e0 \u0111\u00e1ng tin c\u1eady h\u01a1n \u0111\u1ec3 truy xu\u1ea5t v\u00e0 t\u01b0\u01a1ng t\u00e1c v\u1edbi ki\u1ebfn th\u1ee9c c\u1ee7a m\u00ecnh. \n C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u00e3 ch\u1ee9ng minh l\u00e0 m\u1ed9t trong nh\u1eefng c\u00e1ch th\u00e0nh c\u00f4ng nh\u1ea5t, v\u1edbi AI c\u00f3 th\u1ec3 cung c\u1ea5p k\u1ebft qu\u1ea3 t\u1ed1t h\u01a1n tr\u00ean nhi\u1ec1u truy v\u1ea5n. Ch\u00eca kh\u00f3a th\u00e0nh c\u00f4ng n\u00e0y l\u00e0 t\u00edch h\u1ee3p t\u1ea5t c\u1ea3 ki\u1ebfn th\u1ee9c v\u00e0o m\u1ed9t h\u1ec7 th\u1ed1ng th\u1ed1ng nh\u1ea5t, cho ph\u00e9p AI khai th\u00e1c to\u00e0n b\u1ed9 n\u1ed9i dung khi tr\u1ea3 l\u1eddi c\u00e1c c\u00e2u h\u1ecfi. \n Nh\u01b0 \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc, t\u00f4i kh\u00f4ng th\u1ec3 ch\u1ea1y n\u00f3 tr\u00ean trang web c\u1ee7a m\u00ecnh v\u00e0 n\u00f3 ch\u1ec9 kh\u1ea3 d\u1ee5ng v\u1edbi nh\u1eefng ng\u01b0\u1eddi \u0111\u0103ng k\u00fd tr\u1ea3 ph\u00ed c\u1ee7a Chat GPT m\u00e0 t\u00f4i c\u1ea3m th\u1ea5y qu\u00e1 h\u1ea1n ch\u1ebf. Ngo\u00e0i ra, m\u1eb7c d\u00f9 n\u00f3 t\u1ed1t h\u01a1n, t\u00f4i v\u1eabn kh\u00f4ng th\u00edch ch\u1ea5t l\u01b0\u1ee3ng c\u00e2u tr\u1ea3 l\u1eddi v\u00e0 kh\u00f4ng tho\u1ea3i m\u00e1i khi ph\u00e1t h\u00e0nh n\u00f3 cho c\u00f4ng ch\u00fang. \n Tinh ch\u1ec9nh cu\u1ed1i c\u00f9ng: Tr\u1ee3 l\u00fd GPT s\u1eed d\u1ee5ng Model 4o\n M\u1ea3nh gh\u00e9p cu\u1ed1i c\u00f9ng c\u1ee7a c\u00e2u \u0111\u1ed1 trong qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n Fabrice AI xu\u1ea5t hi\u1ec7n c\u00f9ng v\u1edbi s\u1ef1 ra \u0111\u1eddi c\u1ee7a GPT Assistants s\u1eed d\u1ee5ng Model 4o. C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y \u0111\u1ea1i di\u1ec7n cho \u0111\u1ec9nh cao c\u1ee7a m\u1ecdi th\u1ee9 t\u00f4i \u0111\u00e3 h\u1ecdc \u0111\u01b0\u1ee3c trong su\u1ed1t d\u1ef1 \u00e1n. B\u1eb1ng c\u00e1ch s\u1eed d\u1ee5ng c\u01a1 s\u1edf d\u1eef li\u1ec7u vector v\u00e0 tinh ch\u1ec9nh c\u00e1c l\u1eddi nh\u1eafc, t\u00f4i \u0111\u1eb7t m\u1ee5c ti\u00eau \u0111\u1ea1t \u0111\u01b0\u1ee3c m\u1ee9c \u0111\u1ed9 ch\u00ednh x\u00e1c v\u00e0 hi\u1ec3u bi\u1ebft ng\u1eef c\u1ea3nh cao nh\u1ea5t c\u00f3 th\u1ec3 trong c\u00e1c ph\u1ea3n h\u1ed3i c\u1ee7a AI. \n Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y bao g\u1ed3m vi\u1ec7c t\u1ea3i l\u00ean t\u1ea5t c\u1ea3 ki\u1ebfn th\u1ee9c t\u00f4i \u0111\u00e3 t\u00edch l\u0169y \u0111\u01b0\u1ee3c v\u00e0o c\u01a1 s\u1edf d\u1eef li\u1ec7u vector, sau \u0111\u00f3 \u0111\u01b0\u1ee3c s\u1eed d\u1ee5ng l\u00e0m n\u1ec1n t\u1ea3ng cho c\u00e1c t\u01b0\u01a1ng t\u00e1c c\u1ee7a AI. C\u01a1 s\u1edf d\u1eef li\u1ec7u vector cho ph\u00e9p AI th\u1ef1c hi\u1ec7n c\u00e1c t\u00ecm ki\u1ebfm ph\u1ee9c t\u1ea1p h\u01a1n, truy xu\u1ea5t th\u00f4ng tin d\u1ef1a tr\u00ean \u00fd ngh\u0129a ng\u1eef ngh\u0129a c\u1ee7a c\u00e1c truy v\u1ea5n thay v\u00ec ch\u1ec9 d\u1ef1a v\u00e0o vi\u1ec7c kh\u1edbp t\u1eeb kh\u00f3a. \u0110i\u1ec1u n\u00e0y \u0111\u00e1nh d\u1ea5u m\u1ed9t b\u01b0\u1edbc ti\u1ebfn \u0111\u00e1ng k\u1ec3 so v\u1edbi c\u00e1c ph\u01b0\u01a1ng ph\u00e1p ti\u1ebfp c\u1eadn tr\u01b0\u1edbc \u0111\u00e2y, cho ph\u00e9p AI hi\u1ec3u v\u00e0 ph\u1ea3n h\u1ed3i t\u1ed1t h\u01a1n c\u00e1c c\u00e2u h\u1ecfi ph\u1ee9c t\u1ea1p, nhi\u1ec1u s\u1eafc th\u00e1i. \n M\u1ed9t trong nh\u1eefng c\u1ea3i ti\u1ebfn ch\u00ednh c\u1ee7a ph\u01b0\u01a1ng ph\u00e1p n\u00e0y l\u00e0 s\u1ef1 tinh ch\u1ec9nh c\u1ea9n th\u1eadn c\u00e1c l\u1eddi nh\u1eafc. B\u1eb1ng c\u00e1ch t\u1ec9 m\u1ec9 t\u1ea1o ra v\u00e0 th\u1eed nghi\u1ec7m c\u00e1c l\u1eddi nh\u1eafc kh\u00e1c nhau, t\u00f4i c\u00f3 th\u1ec3 h\u01b0\u1edbng d\u1eabn AI cung c\u1ea5p c\u00e1c c\u00e2u tr\u1ea3 l\u1eddi ch\u00ednh x\u00e1c v\u00e0 ph\u00f9 h\u1ee3p h\u01a1n. \u0110i\u1ec1u n\u00e0y kh\u00f4ng ch\u1ec9 bao g\u1ed3m vi\u1ec7c \u0111i\u1ec1u ch\u1ec9nh c\u00e1ch di\u1ec5n \u0111\u1ea1t c\u1ee7a c\u00e1c l\u1eddi nh\u1eafc m\u00e0 c\u00f2n th\u1eed nghi\u1ec7m c\u00e1c c\u00e1ch kh\u00e1c nhau \u0111\u1ec3 c\u1ea5u tr\u00fac c\u00e1c truy v\u1ea5n nh\u1eb1m \u0111\u01b0a ra c\u00e1c ph\u1ea3n h\u1ed3i t\u1ed1t nh\u1ea5t c\u00f3 th\u1ec3. K\u1ebft qu\u1ea3 th\u1eadt \u1ea5n t\u01b0\u1ee3ng. AI hi\u1ec7n c\u00f3 th\u1ec3 x\u1eed l\u00fd nhi\u1ec1u lo\u1ea1i truy v\u1ea5n v\u1edbi \u0111\u1ed9 ch\u00ednh x\u00e1c cao, ngay c\u1ea3 khi c\u00e1c c\u00e2u h\u1ecfi m\u1edf ho\u1eb7c \u0111\u00f2i h\u1ecfi hi\u1ec3u bi\u1ebft s\u00e2u s\u1eafc v\u1ec1 b\u1ed1i c\u1ea3nh. V\u00ed d\u1ee5, khi \u0111\u01b0\u1ee3c h\u1ecfi “L\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 \u0111\u01b0a ra nh\u1eefng quy\u1ebft \u0111\u1ecbnh quan tr\u1ecdng nh\u1ea5t trong cu\u1ed9c s\u1ed1ng c\u1ee7a b\u1ea1n?”, AI \u0111\u00e3 \u0111\u01b0a ra c\u00e2u tr\u1ea3 l\u1eddi to\u00e0n di\u1ec7n v\u00e0 s\u00e2u s\u1eafc, d\u1ef1a tr\u00ean nhi\u1ec1u ngu\u1ed3n v\u00e0 g\u00f3c nh\u00ecn kh\u00e1c nhau \u0111\u1ec3 \u0111\u01b0a ra ph\u1ea3n h\u1ed3i to\u00e0n di\u1ec7n. \n Th\u00e0nh c\u00f4ng n\u00e0y l\u00e0 \u0111\u1ec9nh cao c\u1ee7a h\u00e0ng tr\u0103m gi\u1edd l\u00e0m vi\u1ec7c v\u00e0 v\u00f4 s\u1ed1 th\u1eed nghi\u1ec7m. N\u00f3 ch\u1ee9ng minh r\u1eb1ng, v\u1edbi s\u1ef1 k\u1ebft h\u1ee3p \u0111\u00fang \u0111\u1eafn gi\u1eefa c\u00f4ng ngh\u1ec7 v\u00e0 s\u1ef1 tinh ch\u1ec9nh, c\u00f3 th\u1ec3 t\u1ea1o ra m\u1ed9t AI kh\u00f4ng ch\u1ec9 c\u00f3 th\u1ec3 l\u01b0u tr\u1eef v\u00e0 truy xu\u1ea5t th\u00f4ng tin hi\u1ec7u qu\u1ea3 m\u00e0 c\u00f2n t\u01b0\u01a1ng t\u00e1c v\u1edbi th\u00f4ng tin theo c\u00e1ch c\u00f3 \u00fd ngh\u0129a. S\u1ef1 ph\u00e1t tri\u1ec3n c\u1ee7a GPT Assistants s\u1eed d\u1ee5ng Model 4o \u0111\u00e1nh d\u1ea5u th\u1eddi \u0111i\u1ec3m m\u00e0 Fabrice AI th\u1ef1c s\u1ef1 ph\u00e1t huy h\u1ebft kh\u1ea3 n\u0103ng c\u1ee7a m\u00ecnh, \u0111\u1ea1t \u0111\u01b0\u1ee3c m\u1ee9c \u0111\u1ed9 tinh vi v\u00e0 ch\u00ednh x\u00e1c m\u00e0 t\u00f4i \u0111\u00e3 h\u00ecnh dung ngay t\u1eeb \u0111\u1ea7u. API GPT Assistants sau \u0111\u00f3 \u0111\u00e3 \u0111\u01b0\u1ee3c t\u00edch h\u1ee3p v\u00e0o blog c\u1ee7a t\u00f4i \u0111\u1ec3 cho ph\u00e9p ng\u01b0\u1eddi d\u00f9ng cu\u1ed1i t\u01b0\u01a1ng t\u00e1c v\u1edbi Fabrice AI theo c\u00e1ch b\u1ea1n th\u1ea5y tr\u00ean blog ngay b\u00e2y gi\u1edd. \n Suy ng\u1eabm v\u1ec1 h\u00e0nh tr\u00ecnh\n Qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n Fabrice AI \u0111\u00e3 l\u00e0m n\u1ed5i b\u1eadt s\u1ef1 ph\u1ee9c t\u1ea1p khi l\u00e0m vi\u1ec7c v\u1edbi AI, \u0111\u1eb7c bi\u1ec7t l\u00e0 khi n\u00f3i \u0111\u1ebfn vi\u1ec7c hi\u1ec3u v\u00e0 ng\u1eef c\u1ea3nh h\u00f3a th\u00f4ng tin. N\u00f3 d\u1ea1y t\u00f4i r\u1eb1ng kh\u00f4ng c\u00f3 l\u1ed1i t\u1eaft n\u00e0o trong qu\u00e1 tr\u00ecnh ph\u00e1t tri\u1ec3n AI\u2014m\u1ed7i b\u01b0\u1edbc, m\u1ed7i l\u1ea7n l\u1eb7p l\u1ea1i v\u00e0 m\u1ed7i th\u1eed nghi\u1ec7m \u0111\u1ec1u l\u00e0 m\u1ed9t ph\u1ea7n c\u1ea7n thi\u1ebft c\u1ee7a h\u00e0nh tr\u00ecnh h\u01b0\u1edbng t\u1edbi vi\u1ec7c t\u1ea1o ra th\u1ee9 g\u00ec \u0111\u00f3 th\u1ef1c s\u1ef1 hi\u1ec7u qu\u1ea3. \n Nh\u00ecn v\u1ec1 ph\u00eda tr\u01b0\u1edbc, t\u00f4i r\u1ea5t h\u00e0o h\u1ee9ng \u0111\u01b0\u1ee3c ti\u1ebfp t\u1ee5c tinh ch\u1ec9nh v\u00e0 m\u1edf r\u1ed9ng Fabrice AI. Nh\u01b0 \u0111\u00e3 \u0111\u1ec1 c\u1eadp trong b\u00e0i \u0111\u0103ng tr\u01b0\u1edbc, t\u00f4i s\u1ebd xem x\u00e9t c\u00e1c c\u00e2u h\u1ecfi \u0111\u01b0\u1ee3c h\u1ecfi \u0111\u1ec3 ho\u00e0n thi\u1ec7n c\u01a1 s\u1edf ki\u1ebfn th\u1ee9c khi c\u00f2n thi\u1ebfu s\u00f3t. T\u00f4i c\u0169ng hy v\u1ecdng cu\u1ed1i c\u00f9ng s\u1ebd ph\u00e1t h\u00e0nh m\u1ed9t phi\u00ean b\u1ea3n t\u01b0\u01a1ng t\u00e1c tr\u00f4ng v\u00e0 nghe gi\u1ed1ng t\u00f4i m\u00e0 b\u1ea1n c\u00f3 th\u1ec3 n\u00f3i chuy\u1ec7n. \n ","Category":["Suy ngh\u0129 c\u00e1 nh\u00e2n","Ti\u1ec7n \u00edch c\u00f4ng ngh\u1ec7"],"_links":{"self":[{"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/posts\/46237","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/users\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/comments?post=46237"}],"version-history":[{"count":1,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/posts\/46237\/revisions"}],"predecessor-version":[{"id":46239,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/posts\/46237\/revisions\/46239"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/media\/46234"}],"wp:attachment":[{"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/media?parent=46237"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/categories?post=46237"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/grinda.org\/vi\/wp-json\/wp\/v2\/tags?post=46237"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}