Prompt share: Cinematic portrait
Prompt:
Cinematic portrait of a [subject], [background], soft ambient lighting, warm earthy tones, nostalgic 1970s wardrobe, reflective mood, gentle film grain texture, shallow depth of field, vintage editorial photography style.
Prompt:
Cinematic portrait of a [subject], [background], soft ambient lighting, warm earthy tones, nostalgic 1970s wardrobe, reflective mood, gentle film grain texture, shallow depth of field, vintage editorial photography style.
β€36π₯4
πVisual-language assistant using DeepSeek-VL2 and OpenVINO
π DeepSeek-VL2 is an advanced series of large Mixture-of-Experts (MoE) Vision-Language models.
βοΈ DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.
π° More details can be found in the paper and original repository.
π° In this tutorial we consider how to convert and run DeepSeek-VL2 models using OpenVINO and optimize it using NNCF.
π° Table of contents:
πΉPrerequisites
πΉConvert model to OpenVINO πΉIntermediate Representation
πΉCompress model weights to 4-bit
πΉPrepare inference pipeline
πΉRun model inference
πΉVisual Grounding
πΉGrounding Conversation
πΉVisual Question Answering
πΉInteractive demo
πLinks:
https://github.com/openvinotoolkit/openvino_notebooks/blob/latest/notebooks%2Fdeepseek-vl2%2Fdeepseek-vl2.ipynb
π DeepSeek-VL2 is an advanced series of large Mixture-of-Experts (MoE) Vision-Language models.
βοΈ DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.
π° More details can be found in the paper and original repository.
π° In this tutorial we consider how to convert and run DeepSeek-VL2 models using OpenVINO and optimize it using NNCF.
π° Table of contents:
πΉPrerequisites
πΉConvert model to OpenVINO πΉIntermediate Representation
πΉCompress model weights to 4-bit
πΉPrepare inference pipeline
πΉRun model inference
πΉVisual Grounding
πΉGrounding Conversation
πΉVisual Question Answering
πΉInteractive demo
πLinks:
https://github.com/openvinotoolkit/openvino_notebooks/blob/latest/notebooks%2Fdeepseek-vl2%2Fdeepseek-vl2.ipynb
β€11
Please open Telegram to view this post
VIEW IN TELEGRAM
β€2
Please open Telegram to view this post
VIEW IN TELEGRAM
β€1
Ex_Files_GPT_Models_Tips_Tricks.zip
25.3 KB
Please open Telegram to view this post
VIEW IN TELEGRAM
β€1
From ChatGPT to search to translations β itβs all the same pipeline: data β training β feedback β real-time predictions.
Please open Telegram to view this post
VIEW IN TELEGRAM
β€7π1
π 17,000 prompts in one database - everything you need to work with AI is collected!
The developers have collected a huge repository of queries for all top neural networks: from Midjourney and ChatGPT to Runway and DALL E.
β What's inside:
β’ All prompts are conveniently sorted by categories, tasks, styles and tools - you wonβt get lost.
β’ Usage examples are included with each request.
β’ The service helps to adapt your own prompts to specific tasks.
β’ You can publish your prompts and share them with others.
β’ There is a quick extension for Chrome.
β’ And all this is free.
https://promptport.ai/
The developers have collected a huge repository of queries for all top neural networks: from Midjourney and ChatGPT to Runway and DALL E.
β What's inside:
β’ All prompts are conveniently sorted by categories, tasks, styles and tools - you wonβt get lost.
β’ Usage examples are included with each request.
β’ The service helps to adapt your own prompts to specific tasks.
β’ You can publish your prompts and share them with others.
β’ There is a quick extension for Chrome.
β’ And all this is free.
https://promptport.ai/
β€27π₯11π1
Please open Telegram to view this post
VIEW IN TELEGRAM
β€14
Please open Telegram to view this post
VIEW IN TELEGRAM
β€4
Prompt:
An X-ray scan of a [subject], bones and internal structures glowing in neon cyan and soft magenta against a matte black field. Scientific overlay graphics and scan metadata give it a medical sci-fi twist.
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
π11β€9
Not all AI tools are built the same. Each one shines in different situations.
ChatGPT
A strong all-around assistant for creativity, writing, brainstorming, and everyday problem-solving. Best for content creation, coding help, automation ideas, and fast answers.
Gemini
Googleβs AI is designed to work seamlessly with Gmail, Docs, Drive, and Search. Best for summarizing emails, working with files, research, and multimodal tasks.
Claude
Built for deep reasoning, long documents, and careful analysis. Best for legal work, research-heavy tasks, large codebases, and long-form writing.
Quick rule of thumb:
Creative and versatile β ChatGPT
Google ecosystem and research β Gemini
Long documents and precision β Claude
There is no single βbestβ AI. The right choice depends on what you need to get done.
Save this post. Share it with someone exploring AI.
Please open Telegram to view this post
VIEW IN TELEGRAM
β€24π14π₯2
OpenAI says ChatGPT now has over 800 million weekly users. Thatβs around 10 percent of the worldβs adult population, just three years after launch.
New research shows how people actually use it. More than 70% of conversations are now non-work related, up from 53% early on. When it is used for work, writing dominates, far ahead of search, coding, or analysis.
Nick Turley, head of ChatGPT, put it simply: βThat leaves at least 90 percent to go.β
Please open Telegram to view this post
VIEW IN TELEGRAM
β€21
Please open Telegram to view this post
VIEW IN TELEGRAM
β€7π2
Please open Telegram to view this post
VIEW IN TELEGRAM
β€6
Prompt:
Hyper-realistic black and white sketch of a [subject], detailed pencil linework with subtle graphite shading, accented by a single bold stroke of red. High-resolution, clean composition, fine art illustration style, ideal for gallery wall prints or modern posters.
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
π12β€10