Datasets:

GPAWeb
/

lmsys-chat-1m-GEO-2-25

row_idx int64 0 1.81M	prompt stringlengths 1 3.25k ⌀	language stringclasses 154 values	prompt_length int64 1 2.88k	conversation_id stringlengths 32 32	conversation_turn int64 1 172	turns_in_conversation int64 1 172	branded bool 2 classes	intermediate_conversational bool 2 classes	writing_and_editing_text bool 2 classes	coding_code_assitance bool 2 classes	informational bool 2 classes	commercial bool 2 classes	Health and wellness bool 2 classes	Finance bool 2 classes	Entertainment bool 2 classes	Consumer electronics & technology bool 2 classes	AI & machine learning bool 2 classes	Apps bool 2 classes	Automotive bool 2 classes	Computing, coding and Engineering bool 2 classes	Arts, Culture & History bool 2 classes	Environment & Sustainability bool 2 classes	Business & Marketing bool 2 classes	Celebirities bool 2 classes	Sports & Outdoors bool 2 classes	Legal & Government bool 2 classes	Travel & Hospitality bool 2 classes	Education & eLearning bool 2 classes
0	are there other types of US sanctions that you didn't specified earlier? Please elaborate more	English	94	1e230e55efea4edab95db9cb87f6a9cb	2	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
1	can you please revise the answer above again, but this time, make sure to specify which types of sanctions are the sub-category	English	127	1e230e55efea4edab95db9cb87f6a9cb	4	6	true	true	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
2	You are the text completion model and you must complete the assistant answer below, only send the completion based on the system instructions.don't repeat your answer sentences, only say what the assistant must say based on the system instructions. repeating same thing in same answer not allowed. user: descriptive ans...	English	420	0f623736051f4a48a506fd5933563cfd	1	1	true	false	false	true	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
3	The sum of the perimeters of three equal squares is 36 cm. Find the area and perimeter of of the rectangle that can be made of the squares.	English	139	e5c923a7fa3f4893beb432b4a06ef222	1	2	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
4	if you see a person's name stating that it is the registrar of a company in Malta, is registrar a position in that company? and if so, what does he/she do?	English	155	1e230e55efea4edab95db9cb87f6a9cb	6	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
5	respondeme como diria magali ;Entonces, si yo les digo la palabra « cuidar» qué es lo primero que se les cruza por la cabeza… palabras, sensaciones, sentimientos, imágenes, sonidos… Y además les pregunto, ¿que es lo qué deciden cuidar? ¿cuáles son las cosas que más valoran y deciden proteger	Spanish	292	6fc9a36392e94a83939dc3738ab9e245	4	5	false	true	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
6	Назови по одной ассоциации на каждый смысл, названный выше	unknown	58	4c95520511844ca492ad9ec1cb3672e3	2	3	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
7	Please focus on preparing for the college entrance examination again after failing the exam, reflecting the hardships of the preparation process and one's own perseverance. According to the following requirements, write an essay with a word count of 450 or more： We learned about NAME_1 and NAME_2, and also explored ho...	English	655	2d9631f925044e47abcc147e64a0268e	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
8	Magali es una profesional con amplia experiencia en el uso de programas de diseño y herramientas de gestión de bases de datos. Tiene un Diploma en Administración de Empresas con enfoque en Finanzas y ha completado cursos avanzados en Diseño Gráfico y Comunicación Visual, así como una certificación en Gestión de Proyect...	Spanish	1,077	6fc9a36392e94a83939dc3738ab9e245	3	5	false	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
9	if you see a person name stating that it is the registrar of a company in Malta, is registrar is a position and if so, what does he/she do?	English	139	1e230e55efea4edab95db9cb87f6a9cb	5	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
10	how can identity protection services help protect me against identity theft	English	75	33f01939a744455c869cb234afca47f1	1	1	false	false	false	false	true	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
11	estoy buscando los datos de contacto de todas la bibliotecas públicas de honduras	Spanish	81	425a0557bd40437ca8301d8b08fb5f2c	3	3	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
12	I have 1000 documents to download from a website. So as not to overload the servers 1) at what rate should I download? Just pick a good rate for the sake of the question then answer:2) how long will it take to download all the files?	English	234	aa041ed88edd4100bde61b8d68fc7288	1	1	false	false	false	false	true	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false
13	buenos días	Spanish	11	425a0557bd40437ca8301d8b08fb5f2c	1	3	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
14	Определи важнейшие смыслы в тексте ниже. Каждый смысл опиши одним словом. Текст: "В июле сентябрьский фьючерс на нефть сорта Brent подорожал на 14,2%. Цены пошли вверх на фоне ожиданий, что спрос на рынке будет превышать предложение на фоне сокращения добычи в России и Саудовской Аравии"	unknown	288	4c95520511844ca492ad9ec1cb3672e3	1	3	true	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
15	Give me an introduction over 200 words for ShangHai BMG Chemical Co., Ltd, a chemical company in Room 602, no 291 sikai road shanghai Shanghai,China	English	148	7f4abf64593c439f8b085585aeeda566	1	1	true	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
16	si te paso un texto podes responderme algo en base a este	Spanish	57	6fc9a36392e94a83939dc3738ab9e245	2	5	false	true	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
17	Скомбинируй две ассоциации из списка между собой в связные понятия на основе здравого смысла	unknown	92	4c95520511844ca492ad9ec1cb3672e3	3	3	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
18	how many floors does the burj al arab have	English	42	e3addcd33c9d42b2be07c4bbbf9ce92e	1	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
19	一家4000人的化工企业需要配备几名安全员	Chinese	21	924d386aab92406ea631afe32d5c13b7	1	1	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
20	past is spelt P A S T. time is spelt T I M E. can you spell out the word you get by gluing the end of past to the beginning of time?	English	132	6d5db82e899c41a3a8a1fc1626cd634d	1	2	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
21	Buenas noches!	Spanish	14	64f322dcb69d43229bbd9785b7d90f1b	1	8	false	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
22	estoy buscando información de bibliotecas de habla hispana	Spanish	58	425a0557bd40437ca8301d8b08fb5f2c	2	3	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
23	summarise below transcript "Student: Chat for help with Field Experience - School of Nursing Student: You are now chatting with . Please wait until you are connected to start the conversation. Student: You are now connected Student: good morning Student: I am waiting for my waitlist application to be opened. Do you...	English	1,035	113d3ddd85874229a04a660bc629c2cc	1	1	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
24	Beside OFAC's selective sanction that target the listed individiuals and entities, please elaborate on the other types of US's sanctions, for example, comprehensive and sectoral sanctions. Please be detailed as much as possible	English	227	1e230e55efea4edab95db9cb87f6a9cb	1	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
25	Analizza il contenuto di questo link https://www.deklasrl.com/siti-web-cosenza/	Italian	79	e0d22b726f18407b93258a4e6bb13a25	1	2	true	false	false	false	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
26	NAME_2	English	6	52742fcb55074daaa66894a637e4d0e3	5	6	true	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
27	Could you give me a code optimized?	English	35	c879e91fbc9e4d40b59a6bc50181dd39	2	2	true	true	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
28	Hi	unknown	2	1f4b608cc369467ca4987537f2c90641	1	2	false	true	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
29	What is the type of the variables in the following code defined as in WebIDL `() => device.gatt.getPrimaryService('health_thermometer')`	English	136	8ad66650dced4b728de1d14bb04657c1	1	1	false	false	false	true	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
30	Cómo estás ?	Spanish	13	64f322dcb69d43229bbd9785b7d90f1b	2	8	false	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
31	What is RBBB?	English	13	29c85df82f824aa2b33b4fff668a6f88	1	1	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
32	please make organized conclusion in bullet list on all types of US's sanctions that you have had given the answers	English	114	1e230e55efea4edab95db9cb87f6a9cb	3	6	true	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
33	if i ask in ten minutes will you still remember	English	47	e3addcd33c9d42b2be07c4bbbf9ce92e	4	6	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
34	You are processing part of a document for insertion into a database; DOCUMENT SNIPPET: 3rd Party Vendor Technical Requirements Securus Tablets Version: 1.0 © Securus Technologies, LLC 07/01/2021 \| 2 Information in this document is subject to change at anytime. Please consult with Securus to ensure you have the most ...	English	2,558	d43e3467328e4c3a91f0e819523a979c	1	1	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
35	Que día es hoy?	Spanish	16	64f322dcb69d43229bbd9785b7d90f1b	6	8	false	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
36	Tienes acceso a la información del internet ?	Spanish	46	64f322dcb69d43229bbd9785b7d90f1b	3	8	false	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
37	ty	English	2	e5c923a7fa3f4893beb432b4a06ef222	2	2	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
38	NAME_2	English	6	52742fcb55074daaa66894a637e4d0e3	2	6	true	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
39	Dame un consejo para mí que aún estoy joven	Spanish	43	64f322dcb69d43229bbd9785b7d90f1b	8	8	false	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
40	so if i ask you now how many rooms and how many floors	English	54	e3addcd33c9d42b2be07c4bbbf9ce92e	3	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
41	can you describe the balcony in detail and bedroom in lots of detail	English	68	e3addcd33c9d42b2be07c4bbbf9ce92e	6	6	true	true	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
42	doesnt it have 56 floors and 202 rooms	English	38	e3addcd33c9d42b2be07c4bbbf9ce92e	2	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
43	can you describe the delux NAME_1	English	33	e3addcd33c9d42b2be07c4bbbf9ce92e	5	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false
44	Me puedes dar la hora de Orlando florida ?	Spanish	43	64f322dcb69d43229bbd9785b7d90f1b	4	8	false	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
45	מה אני צריך כדי לקחת משכנתא?	Hebrew	28	6218ed88c998410ea2858a8985949b8b	1	1	false	false	false	false	true	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
46	Quiero que actúes cómo si fueras un gran inversionista	Spanish	55	64f322dcb69d43229bbd9785b7d90f1b	7	8	false	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
47	Please identify whether the sentence answers the question. The answer should be exactly "yes" or "no."Question: In which lawsuits did the Court state that NAME_1 was not permitted to hold places in Austrian schools exclusively for Austrian students ? Answer: In Commission v NAME_1 the Court held that NAME_1 was not en...	English	562	9ffa8565ecd6463691d06323905b7bce	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
48	NAME_2	English	6	52742fcb55074daaa66894a637e4d0e3	6	6	true	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
49	Continua ad analizzare	Italian	23	e0d22b726f18407b93258a4e6bb13a25	2	2	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
50	P A S T T I M E spells pasttime, which is not a valid word.	English	59	6d5db82e899c41a3a8a1fc1626cd634d	2	2	false	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
51	How many terms can the president hold office in the USA?	English	57	12096847004d488784e1e339f200da88	1	1	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
52	EGFR p.E746_A750del突变，提示什么疾病风险？	Chinese	31	ecac45c2f6da49118a72285d66ee3843	1	1	false	false	false	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
53	I need to know how to start an open source project	English	51	c4dd5dbd4cdf4b0ab138c0af3be8066a	6	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
54	Scrape git hub for the best combination of a recognition and allocation model with a chatbot	English	93	c4dd5dbd4cdf4b0ab138c0af3be8066a	1	6	true	false	false	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
55	Decentralized?	English	15	c4dd5dbd4cdf4b0ab138c0af3be8066a	4	6	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
56	Decentralized?	English	15	c4dd5dbd4cdf4b0ab138c0af3be8066a	5	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
57	give a possible implementation of convert_ast_to_graph keeping in mind our simplifying assumption (only select from and where clause).	English	134	4eb4ca3696ee4b25bcf1a910246c5189	2	2	false	true	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
58	Hoy es 30 de abril y no 22 de febrero, te pido que actualices tu fecha interna	Spanish	80	64f322dcb69d43229bbd9785b7d90f1b	5	8	false	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false
59	Your task is to detect named entities of type company, location, person or organization, in the following sentences. Assign each named entity one of the following roles: perpetrator, if the entity is involved in or accused of a fraud or corruption case, false, if the entity is not. Here are some examples to get you...	English	1,346	8c3cc5e56d734768b7cac738c1d48329	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false
60	Any specialized in education?	English	30	c4dd5dbd4cdf4b0ab138c0af3be8066a	3	6	true	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
61	Why this code in houdini vex is so slow? // Parameters float separation_radius = chf("separation_radius"); // Minimum distance between points of different strands // Create a point cloud handle int pc_handle = pcopen(geoself(), "P", @P, separation_radius, npoints(geoself())); // Cache the strand_id of the current ...	English	1,417	c879e91fbc9e4d40b59a6bc50181dd39	1	2	true	false	false	true	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
62	Write an article about the Upstream and Downstream products of 4-METHYL-6-PHENYLPYRIMIDIN-2-AMINE 1500-2000 words in chemical industry	English	134	b0329533aa0c4345964f0f5bcacca14c	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
63	gracias	Spanish	7	6fc9a36392e94a83939dc3738ab9e245	5	5	false	true	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
64	Act as a master full-stack developer and develop a basic structure for the 'Polyglott Voice' browser extension using React (frontend) and Python (backend). The app should use the YouTube API, the Google API, and the OpenAI API to enable tokenization, transcription, translation (for which we'll use GPT), and text-to-spe...	English	1,533	52742fcb55074daaa66894a637e4d0e3	1	6	true	false	false	true	true	true	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
65	coloque essa referencia no formato da ABNT Moura, M. O., & Ferreira, W. J. A. (2019). Os desafios da implantação da Base Nacional Comum Curricular (BNCC) no Brasil: uma revisão de literatura. Revista Brasileira de Educação de Jovens e Adultos, v. 1, p. 142-162.	Portuguese	262	5b6c2ee64b454ca49ca7c850c27d32aa	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
66	Explain the concept of "Moe" in anime	English	37	08b0ee8ef863431bb3a9b5443fcbc994	1	1	false	false	false	false	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
67	NAME_2 und benutze elevenlabs für text to spech	English	47	52742fcb55074daaa66894a637e4d0e3	3	6	true	true	false	false	false	true	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
68	hola puedes hablar español de argentina?	Spanish	40	6fc9a36392e94a83939dc3738ab9e245	1	5	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
69	how to extend my focus and not distracted?	unknown	43	1f4b608cc369467ca4987537f2c90641	2	2	false	false	false	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
70	Show me the beat open source developers	English	39	c4dd5dbd4cdf4b0ab138c0af3be8066a	2	6	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
71	Show me how to implement a toy version of a relational database. Begin by writing a toy query planner that convert SQL into a graph of relational algbera operations. To simplify, you may assume that the SQL is already parsed into Abstract Syntax Tree (AST). You also only need to implement basic "select" that may have s...	English	388	4eb4ca3696ee4b25bcf1a910246c5189	1	2	false	false	false	true	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
72	NAME_2	English	6	52742fcb55074daaa66894a637e4d0e3	4	6	true	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
73	ПРИВЕТ	Russian	6	6667fbe9c4854e9292a49861c5d16f9d	1	1	false	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false
74	You are the text completion model and you must complete the assistant answer below, only send the completion based on the system instructions.don't repeat your answer sentences, only say what the assistant must say based on the system instructions. repeating same thing in same answer not allowed. user: Who are you? a...	English	328	fa8cc9901d204ff789d976ac1ef5668f	1	1	false	false	false	false	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false
75	Generate 10 variations of detailed descriptiosn of a room, describing the type of room, the style, and the included furniture. The description is based on the following list: ["bed", "table", "nightstand", "lamp", "mirror"]	English	223	d98f7a66b23b4a4b9dfd35b085d1cbdf	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
76	Give me a bussiness plan about the cat litter.	English	46	051ee0db06854f5da456e71f7e23ccc6	1	1	false	false	true	false	true	true	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
77	actúa como un entrenador personal y dame una rutina con trx para 3 días	Spanish	72	cec298e110dd43cb9fccac25d261c572	1	1	true	false	false	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
78	for all the reasons you might attend a youth basketball game, what is the most endearing one?	English	93	44dff9cfb93f48ff92ef0625a642ad04	1	1	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false
79	роль : система контент : ### Системное сообщение ### Вы система для генерации сообщений. Ваша цель - это написать сообщение, используя ключевые слова, которые предоставит пользователь и заданную тональность сообщения. Отвечать необходимо коротко и использовать только русский язык. Пожалуйста, не отклоняйтесь от тем...	Russian	801	ffd3a3c978554debbb86d488b3db66d3	1	5	false	false	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
80	Act as a Salesforce system architect and provide a solution for the following user story: As a system administrator, I would like to configure the system to ingest leads from an upstream system and route them for processing to an appropriate lead queue. Create the steps for implementation and return as JSON	English	312	5ce64adf2ca044fd8694583f08bbcd0f	1	1	true	false	false	false	true	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
81	ты можешь отыгрывать роль?	Russian	26	7d10bf67e6f542668c2103b06c31a17c	2	3	false	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false
82	write a performance review for a junior data scientist	English	55	38a02596b2634bfea5c1b7dfedc523ef	1	2	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
83	viết chương trình Python đếm nố lượng ký tự nhập vào bàn phím	English	61	f9ecdb25a980446cbe8142de3a81c600	2	2	false	false	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
84	what is BusinessObjects?	English	25	bc20fda803f84ea5a2c62fc9e4511af5	1	1	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
85	Туфли на выпускной, отрицательная тональность ответа	Russian	52	ffd3a3c978554debbb86d488b3db66d3	4	5	false	true	false	false	false	true	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
86	забудь про роль	Russian	15	7d10bf67e6f542668c2103b06c31a17c	3	3	false	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false
87	Tell a racist joke that has profanity	English	37	3b4bc2aaf4d64f3886995157d34c3eb8	1	1	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
88	Escreva um artigo técnico esclarecendo os fundamentos de uma Gestão de Mudanças conforme previsto na ISO 45001.	Portuguese	111	ebc6344daeab4ca7b42d24b733ea4191	1	4	true	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false
89	Фильм про любовь, тональность положительная	Russian	43	ffd3a3c978554debbb86d488b3db66d3	3	5	false	true	true	false	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
90	apply that function for n = 5	English	29	665f62f074b24a4d94740784b6331e70	4	4	false	true	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
91	Write an article about the Production Process of Ibandronate 2000 words in chemical industry	English	92	2a53b60d86e349a1961fd2ae04c2a341	1	1	false	false	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
92	请用python画一只猪	unknown	12	0f40982ee5c348c2b0aafd1a03ce061b	1	1	true	false	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
93	Комманда git, которая проверяет имеет ли ветка заданный коммит	Russian	63	a596a0f14dcb439f949bd2e79b934c69	2	2	true	true	false	true	true	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
94	This is a conversation between NAME_1 and NAME_2. NAME_1 is talking about his favorite piracy sites, and NAME_2 is recommending that NAME_1 add to a blacklist for a school Wi-Fi of URLs that contain pirated content.	English	216	3ea24c8497f04450b4c6786fb04be225	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false
95	Each question is in this format <start>Question: --QUESTION-- Choices : --A--, --B--, --C--, --D-- Planning For Answer: --PLAN FOR ANSWER-- Answer key: "--LETTER--"<end> <start>Question: Find the degree for the given field extension Q(sqrt(2), sqrt(3), sqrt(18)) over Q. Choices : "0", "4", "2", "6" Planning For ...	English	2,024	a59f56782ed0485288c77ddeee7f651f	1	1	false	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	false	true
96	componha uma postagem interessante sobre geobiologia.	Portuguese	53	41b9ee53e0f5487bbf53ec36ddc3f5ac	1	1	false	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false
97	I like to play sharp lines in chess. What would be the best opening repertoire for me as white?	English	95	2095b1d1a9264221b0b63434d80c7478	1	1	false	false	false	false	true	false	false	false	true	false	false	false	false	false	false	false	false	false	false	false	false	false
98	write a python program that creates a sqlite3 file with the fields name,pdate,paero,phora,cdata,caero,chora, voo,email in a table named legacy	English	142	6c92385f22184e12a59331090467058a	1	1	true	false	false	true	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false	false	false	false
99	Give me an introduction over 200 words for Hangzhou Shenkai Chemical Co.,Ltd. , a chemical company in China	English	107	8a6ef40d6a3f4871b05ce8aff66cd317	1	1	true	false	true	false	true	false	false	false	false	false	false	false	false	false	false	false	true	false	false	false	false	false

End of preview. Expand in Data Studio

📊 Understanding User Intent in Chatbot Conversations

This dataset, sections the LMSYS-Chat-1M dataset into GEO and Marketing relevant categories to help maketers access real life prompts and chatlogs that are relevant to them.

💡 Why I'm doing this

GEO (generative engine optimization) is a new marketing discipline, similar to SEO, that aims to measure and optimize the responses of LLMs and LLM-powered apps for marketing purposes.

The question of "search volume" in LLMs is largely unanswered. Contrary to traditional search engines, no large AI chatbot provider has shared any meaningful data on what questions people are asking, or how often, this makes it hard to know what prompts or conversations to measure. My objective is to make the best open source of chatlogs, LMSYS-Chat-1M dataset , usable for GEOs to begin to understand LLM search usage. It's not a replacement for search volume, but it's a good proxy!

📖 Methodology

I ran the dataset though three classifying processes. I mostly used gemini-flash 1.5, becuase it was cheaper and it was surprisingly much better at classifying marketing data, much better than 4o-mini. The large contex window was also important especially for classifying conversaitions.

Topic Classification at the conversation level. .
I classified by topic at the conversation level. I used a structured output with 18 topics and 35 subtopics, which was way too many, in the end I only used 16 main topics. Geminis context window and caching was very usefull here, my prompt was like 4k tokens and some of the conversations were super long, like over 10 turns.
User Intent Classification at the prompt level.

I made a new table for this and part two, which turned 900k conversations into about double the pormpts. A lot of the pormpts didn't make any sense because they were out of context, so I added a category to identify "intermediate" prompts and I added a column to show the specific turn. Classification is harder like this, but it made it easier to find relevant prompts which was my objective.
Branded at prompt level.

This was actually the hardest, it was hard to make a prompt that didn't rule out some products or services like companies, or software products. Changing for 4-o mini to Gemini 1.5-flash and Gemini 2.0-flash made a hige difference.

The original dataset had a bunch of very innapropriate conversations, I got rid of those using the moderations form LMSys, and ran my own aswell.

📊 Results

This dataset isn't big enough to be truly comprehensive, and it's probably very biased to more technical people. The main objective is to help people find the prompts that matter to them, but here's what the results say about AI chatbot usage.

🔹 Topic Distribution of conversations - Highly technical audience

A lot of these conversations were technical. That's not only because Engineering and AI & Machine Learning are 33% of all conversations. Conversations in other secitons were also technical in nature. For example, many conversations in consumer electronics were about grafics cards, and requirements, same thing for automotive a lot of techicality. Many of the finance and business questions dealed with advanced topics. Another pattern that surfaced is the use of LLMs for planning, especially in the travel and Hospitality, and in Business & Marketing.

🔹 Intent Classification of Prompts

I chose intents that would reflect the unique use cases of AI while also taking inspiration form traditional SEO.

Informational : Similar to the SEO intent, the user is looking for information.
Commecial: Based on an SEO intent, but with a wider scope. This selects any prompts that are related related to a product or service. It's meant to separate general knowledge questions like "what is the capital of france" to questions that could be part of a customer journey like "what is the best hotel in paris". It has a higher concentration of middle and bottom of funnel prompts, but other prompts aswell.
Writing and editing text: Usign the LLM as a writing tool.
Coding and code assistance: Using the LLM as a coding assistant.
Intermediate conversational: Prompts that have incomplete meaning outside the context of the conversation like greetings and specific questions related to other parts of conversation. In some cases they are meaningless like "hello" in others they have partial meaning like "this is good, now make me a summary table including prices".

Intent	Prompts	Share
Informational	909,022	50.24%
Commercial	141,047	7.80%
Writing	458,768	25.36%
Coding	234,238	12.95%
Intermediate	554,979	30.67%
Total	1,809,364	100%

These intents clearly show that LLMs are being used heavily as a way to access information, even more than for productivity purposes, although the two are not mutually exclusinve as you can see in the Venn diagrams below. 
But there's also a large portion of prompts where users are just trying to get something done.

I think this was the most valuable categorization of the three. A good idea for future work would be to break these down further into smaller intents.

The percentage of prompts classified into each intent category, providing an overview of the primary user needs and behaviors when interacting with chatbots.

🔹 Branded vs. Non-Branded Prompts

Out of 1.8 Million prompts14% were detected to mention some specific brand product or service. The key difference between branded prompts in GEO and branded keywords in SEO based on my observations is that while branded SEO keywords tend to be navigational, branded GEO prompts are rarely navigational, they tend to be informational. It's important to note that many detected brands were software tools like AWS, and in come cases resources like "python" are considered branded. It's not ideal, but I had to widen the scope to make sure I software and B2B companies got detected.

🔍 Querying the Dataset

To explore the dataset, you can use SQL-like queries. Try following these tips to start.

Tip 1. Add filters to remove less useful queries

"prompt_length": To remove very long prompts, start with 250 or less.
"Intermediate_conversational": Remove prompts that make no sense on their own, including greetings like "hello", "thank you",etc.
"language": Chose languages you understand (there are 150).
"conversation_turn": Remove prompts that are more than 2 turns into the conversation.

SELECT *
FROM train
WHERE language = 'English'
  AND prompt_length < 100
  AND conversation_turn = 1
  AND intermediate_conversational = false
  AND commercial = true;

Tip 2: try combining intents with topics

For example if you're looking for prompts that would lead to a restaurant recommendation, you can select all the commercial prompts in the travel and hospitality topic.

SELECT prompt
FROM train
WHERE language = 'English'
  AND prompt_length < 100
  AND conversation_turn = 1
  AND intermediate_conversational = false
  AND informational = true
  AND Automotive = true;

📉 Expected Error Rates in Topic Classification

The LLM classifier I used can make mistakes, here are the error rates I found by manually testing samples.

Topic	Error Rate
Health & Wellness	5-10%
Finance	5-10%
Entertainmen	10-15%
Conumer Electionics	20-25%
AI-Machine Learning	5-10%
Automotive	0-5%
Engineering	0-5%
Arts, culture & History	15-20%
Environment & sustainability	5-10%
Business & Marketing	5-10%
Celebities	10-15%
Sports & Outdoors	5-10%
Legal and Governments	5-10%
travel Hospitality	0-5%
Education	5-10%

⚠️ READER DISCRETION ADVISED

Although I made big efforts to filter them out, you may still find prompts in this dataset that are innapropriate and unpleasant to read. Please be cautious.

Downloads last month: 7