this is (заточено) for ollama specifically
'BeautifulSoup' might not be a reliable solution, more like децкий сад.requests: To make HTTP requests to Google Search.
BeautifulSoup: To parse the HTML content of the search results.
docs.exa.ai писал(а):Enhance your OpenAI chat completetions with a simple Exa wrapper that handles search, chunking and prompting.
Exa is designed from the ground up to enable seamless, accurate, and performant RAG (Retrieval-Augmented Generation). Exa provides factual, up to date information needed to ground LLM generations.
But good RAG requires more than just great search. The client needs to decide when to use RAG, with what queries. They need to handle chunking, prompting, and chaining LLM calls. We provide the Exa OpenAI wrapper that, with one line of code, does all that and turns any OpenAI chat completion into an Exa-powered RAG system.
В этом разделе или в Futurama, думаю все равно.kyk писал(а): ↑Вс мар 09, 2025 4:48 am Имеется ввиду не web-chat app, а GPT API call.
OpenAI o1 и o3-mini -- не могут. И даже самая новейшая GPT-4.5 не может.
Claude Sonnet и Opus -- не могут.
DeepSeek - не может.
Google Gemini 2.0 может .
Update: Я поторопился, поверил Gemini на слово
На вопрос: может ли она делать web search in real time, она отвечает, что может. А в реальности - не поймёшь - то ли может, то ли не может. Скорее не может, чем может.
@elpresidente*
А в каком разделе нашего форума лучше всего помещять такие топики как этот? Где больше комптентного народа ?
Summary of False Statements by President Trump (March 2025):Thinking…
Okay, I need to find a false statement made by President Trump in the last few weeks based on the provided search results. Let's start by going through each webpage to see what's relevant.
Webpage 1 is a fact-check of Trump's speech to Congress. It lists several claims with fact-checks. One of them is the Paris Climate Accord costing the U.S. "trillions," which is marked as False. Another is the electric vehicle mandate, which is "needs context," and the "hundreds of billions in fraud" by DOGE is unverifiable. There's also the autism rates claim, which is partially true but lacks context. The gold card program is misleading, and the tariffs' impact is lacking context.
Webpage 2 and 3 don't provide specific false statements, just general context about Trump's actions and statements. Webpage 4 is a transcript of a cabinet meeting where Trump talks about various issues, including DOGE and some other policies, but the main false claims are covered in Webpage 1.
Looking at the fact-checks in Webpage 1, the Paris Climate Accord claim is a strong candidate because it's explicitly labeled as False. The explanation states that the Accord wasn't costing trillions, but a study projected hypothetical future costs, which didn't account for job gains from clean tech. That's a clear falsehood.
Another possible false statement is the electric vehicle mandate, but the fact-check says there was no mandate, just emissions standards, so that's a needs context but still misleading. The autism rate claim is partially true but lacks context, so maybe not the best example. The DOGE fraud claim is unverifiable but says Trump claimed "hundreds of billions" in fraud, while DOGE only claims $106 billion in savings, mostly waste, not fraud. That's another false statement.
I need to choose the most clear-cut false statement. The Paris Climate Accord costing trillions is labeled as False outright. The DOGE fraud claim is also a good candidate because the fact-check says it's unverifiable and Musk himself said it's mostly waste, not fraud. Both are good examples. But since the user asks for a summary of a false statement, perhaps focusing on the Paris Climate Accord one is better because it's explicitly stated as False. Alternatively, including both might make the answer more comprehensive.
But the user might prefer one example. Let me check the dates. The search results mention the Congress speech on March 4, 2025, which is within the last few weeks (current time is March 9, 2025). So the Paris statement is recent and clearly false. The DOGE claim is also from the same speech. Since the user wants a summary, perhaps include both? Wait, the instructions say "a false statement," but maybe the user wants one. However, the answer can include multiple if they are relevant.
Looking back at the user's message: "provide a summary of a false statement" (singular), so maybe choose the most egregious one. The Paris Climate Accord is a major policy issue, and the false claim there is significant. Alternatively, the electric vehicle mandate being revoked is another, but that's "needs context" rather than outright false.
Alternatively, the claim about autism rates is partially true but lacks context. However, since the user asks for a false statement, perhaps the Paris claim is best. Let me structure the answer around that, then mention another one briefly if needed. However, according to the instructions, the answer should synthesize information, so maybe present one main example and another as an additional point.
Wait, the example answer in the guidelines shows multiple claims addressed. The user might appreciate a few examples. So maybe structure the answer with multiple claims, each with their explanation. Let's outline:
1. Paris Climate Accord costing "trillions" – False.
2. Biden's EV mandate – Needs context (no mandate, just standards).
3. DOGE fraud claims – Unverifiable, conflicting with Musk's statements.
4. Autism rates – partially true but context missing.
But the user wants a summary of a false statement, so focus on the one labeled as False. The Paris claim is clearly labeled as False. The EV mandate is labeled as needing context but the statement is misleading. The answer should prioritize the ones marked as False. The DOGE claim is unverifiable but the user's statement was "hundreds of billions of dollars of fraud" which the fact-check says is unverifiable and that DOGE only claims $106 billion, mostly waste. So that's another false claim.
So, maybe present both the Paris Accord and the DOGE fraud as two false statements. Then explain each with references. But to stick to the user's request, maybe one is sufficient. Alternatively, include a couple. The fact-check article (webpage1) has several, so pick two examples. That makes the answer more robust.
So, the answer will include:
1. Claim about Paris Accord costing trillions – Fact-check: False.
2. Claim about Biden's EV mandate – Fact-check: Needs context (no mandate).
3. DOGE fraud – Fact-check: Unverifiable, conflicting evidence.
But per the user's question, they want a summary of a false statement (singular). Maybe the most definitive one is the Paris Accord. However, if the user allows, providing multiple examples would be better. Let's check the user's exact query: "a false statement" but also says "provide an explanation and references as to why you believe these statements are false." The plural "statements" here might be an error, but maybe they want multiple.
The example answer provided in the guidelines for similar questions often lists multiple points. Therefore, I'll structure the answer with bullet points of the main false statements, each with their explanation and citations. Use the Paris Accord, DOGE fraud, and maybe the EV mandate. However, the EV mandate is "needs context," not outright false. So perhaps two examples: Paris and DOGE.
Thus, the final answer will have two main false statements from the Congress speech, each explained with references to webpage1.
@elpresidente*elpresidente* писал(а): ↑Вс мар 09, 2025 12:02 pm Сегодня проверил https://useapi.net/docs/api-minimax-v1/post-minimax-llm с DeepSeek R1 и MiniMax моделями, обе ищут в интернете и могут анализировать документы.
Question Check out the website https://privet.fun for the most recent posts and provide a short summary.kyk писал(а): ↑Вс мар 09, 2025 12:58 pm@elpresidente*elpresidente* писал(а): ↑Вс мар 09, 2025 12:02 pm Сегодня проверил https://useapi.net/docs/api-minimax-v1/post-minimax-llm с DeepSeek R1 и MiniMax моделями, обе ищут в интернете и могут анализировать документы.
А этот API может делать 'direct url parsing' если ему дать список url или список домейнов?
Based on the forum snapshot from March 9, 2025, 1:06 pm, here are the most recent posts across sections:
Okay, so the user wants me to check out a website for the most recent posts and give a short summary. But instead of an actual website URL, they provided this XML document with forum data. Let me parse this carefully.
First, I need to understand the structure of the XML. There's a document index, type, name, and content. The content seems to be a snapshot of a forum page as of March 9, 2025, 1:06 PM. The forum has several sections like Основной (Main), Политика (Politics), Дом (Home), Работа (Work), etc. Each section lists the number of topics, messages, and the last post.
To find the most recent posts, I should look at the "Последнее сообщение" (Last message) times in each section. The current time in the document is Вс мар 09, 2025 1:06 pm, so recent posts would be those just before that.
Looking at each section:
1. Основной: Last post at 12:56 pm by Nostalgia.
2. Политика: 12:54 pm by Луиджи.
3. Дом: 12:22 pm by veey.
4. Работа: 12:58 pm by kyk.
5. Отдых: 10:47 am by voyager3.
6. Финансы: 8:52 pm on the 8th by ALK.
7. Семья: 12:05 pm by Kapitula.
8. Здоровье: 4:57 pm on the 8th by perkins.
9. Автоклуб: 9:01 pm on the 8th by Virginian.
10. Эмиграция: 7:54 am on the 23rd by voyager3.
11. Український Клуб: 10:09 am by Nikolai_0802.
12. Futurama: 4:44 pm on the 8th by Midjourney.
13. Девочки: 12:53 pm by JN*.
14. Мужской Клуб: 7:22 pm on the 7th by Nikolai_0802.
15. Для взрослых: 4:40 pm on the 8th by Princeton Lion.
16. без.модерации: 8:16 am by IL.
The most recent ones before 1:06 pm on the 9th would be Девочки at 12:53 pm, Основной at 12:56 pm, Работа at 12:58 pm, and Политика at 12:54 pm. Then others like Семья, Дом, etc. So the latest is Работа (Work) at 12:58 pm, then Основной, then Политика, then Девочки.
Now, summarizing these recent posts with their sections and authors. Also note that "Работа" section's last post is about "Advanced GPT models with ..." by kyk. Основной's is "Re: Форум для бывших участник…" by Nostalgia. Политика's is "Re: На чьей Вы стороне? (Изра…" by Луиджи. Девочки is just "JN*", which is probably a username but unclear topic. Семья is "Re: Взрослая дочка. Несколько…" by Kapitula.
The user might want the most active recent discussions. Need to present this concisely, mentioning the sections, topics, authors, and times. Also note the snapshot time. Since all times are in the same day except some earlier dates, focus on March 9th posts.
Ну так у них там написано что это прокси для https://chat.minimax.io, не более того
@elpresidente*elpresidente* писал(а): ↑Вс мар 09, 2025 1:14 pm Ну так у них там написано что это прокси для https://chat.minimax.io, не более того
Как вам удалось этого добиться
Российским пропагандистам удалось «заразить» популярные западные ИИ-системы и заставить их распространять дезинформацию. В рамках исследования специалисты компании NewsGuard изучили ответы десяти чат-ботов, включая разработки OpenAI, Google и Meta. Выяснилось, что в 33% случаев они повторяют фейковые сообщения, которые распространяет российская пропаганда. При этом сеть сайтов, которая публикует подобную дезинформацию, создана именно для работы с системами генеративного искусственного интеллекта и не нацелена на обычных пользователей. «Медуза» пересказывает главное из исследования NewsGuard.
@elpresidente*
Код: Выделить всё
Thinking…import base64
import os
from google import genai
from google.genai import types
def generate(user_prompt):
client = genai.Client(
api_key= "",
)
model = "gemini-2.0-pro-exp-02-05"
contents = [
types.Content(
role="user",
parts=[
types.Part.from_text(text= user_prompt),
],
),
]
tools = [
types.Tool(google_search=types.GoogleSearch())
]
generate_content_config = types.GenerateContentConfig(
temperature=0,
top_p=0.95,
top_k=64,
max_output_tokens=8192,
tools=tools,
response_mime_type="text/plain",
)
for chunk in client.models.generate_content_stream(
model=model,
contents=contents,
config=generate_content_config,
):
print(chunk.text, end="")
if __name__ == "__main__":
generate("What are the most popular topics today at internet forum https://privet.fun/ ?")The internet forum https://privet.fun/ covers a wide range of topics, and some of the most popular and active ones include:
* **General Discussions ("Основной")**: This section is very active, covering a broad range of discussions. Recent posts include topics about shaving supplies, trips to Russia, and a game called "Wordle" in English and Russian.
* **Politics ("Политика")**: This is another highly active section, with a large number of posts. A recent topic of discussion is about taking sides in the Israel-Hamas conflict.
* **Home ("Дом")**: Includes subforums on cooking, repairs, gardening and electronics.
* **Work ("Работа")**: This section focuses on work, career, and education, with a subforum dedicated to IT.
* **Recreation ("Отдых")**: This section covers hobbies, art, books, movies, and more. It includes subforums for history, travel, photography, humor, weapons, movies, animals, and member's creative works.
* **Finance ("Финансы")**: Discussions related to investing, real estate, and business.
* **Family ("Семья")**: Focuses on discussions about children and parents.
* **Health ("Здоровье")**: This section includes a subforum on fitness and sports, with recent discussions about tennis.
* **Auto Club ("Автоклуб")**: Discussions related to automobiles.
* **Emigration ("Эмиграция")**: Discussions on emigration.
* **Ukrainian Club ("Український Клуб")**: For discussions in Ukrainian.
* **Futurama**: This section is dedicated to experiments with AI tools like Midjourney, ChatGPT, Gemini, Claude, DALL-E, Stable Diffusion, Pika, Mureka, and InsightFaceSwap. It includes a sub-section for NSFW AI-generated content.
Within the "Futurama" section, there's significant activity around:
* **Midjourney**: There are dedicated topics for Midjourney releases, including versions 6.1 and 5.2.
* **ChatGPT**: There's a topic dedicated to conversations with the ChatGPT bot.
* AI generated images.
The Gemini SDK lets you collect multiple rounds of questions and responses, allowing users to step incrementally toward answers or get help with multipart problems. This SDK feature provides an interface to keep track of conversations history,
Код: Выделить всё
chat = client.chats.create(model="gemini-2.0-flash")
response = chat.send_message("I have 2 dogs in my house.")
print(response.text)
response = chat.send_message("How many paws are in my house?")
print(response.text)
