Post

Conversation

金のニワトリ
@gosrum
karakuri-vl-32b-instruct-2507をお試し。 ※ぱぷりか炒めさんのQ4_K_M量子化版をollamaで使用させていただきました 試しにスクリーンショットのOCRをしてもらったら、完璧だった！！数式をTex形式で出力もできるし、中国語も混ざってない。 これはかなり素晴らしいモデルなのでは？
The image shows a screenshot of a software interface in Japanese, likely related to AI and OCR functionality. The interface includes options for various AI tasks such as text generation, image generation, and OCR. The post text mentions testing the karakuri-vl-32b-instruct-2507 model, specifically the Q4_K_M quantized version used with ollama, and highlights its perfect OCR performance, ability to output mathematical formulas in Tex format, and handling of mixed languages without issues. The screenshot is used to demonstrate the model's capabilities.
The image shows a screenshot of a software interface in Japanese, likely related to AI and OCR functionality. The interface includes options for various AI tasks such as text generation, image generation, and OCR. The post text mentions testing the karakuri-vl-32b-instruct-2507 model, specifically the Q4_K_M quantized version used with ollama, and highlights its perfect OCR performance, ability to output mathematical formulas in Tex format, and handling of mixed languages without issues. The screenshot is used to demonstrate the model's capabilities.
Quote
ぱぷりか炒め
@WMjjRpISUEt2QZZ
karakuri-aiさんが公開しているkarakuri-vl-32b-instruct-2507のggufフォーマット変換版です imatrixのデータはTFMC/imatrix-dataset-for-japanese-llmを使用して作成しました huggingface.co/mmnga/karakuri
19.6K
Views