@kinumaru0: どしたん話聞こか? #夏目きぬ#恋愛心理 #恋愛#質問返し

夏目きぬ。
夏目きぬ。
Open In TikTok:
Region: JP
Monday 24 November 2025 07:14:50 GMT
41713
419
11
23

Music

Download

Comments

tarakochan0912_super
yoiko🍎🐥🌾🎶🐥🌙 :
きぬと予定共有したい
2025-11-24 12:35:45
2
user665918555450
藤喜正行 :
👍可愛い😙😚😘
2025-11-24 08:23:01
1
kinumaru0
夏目きぬ。 :
みんなは共有してる???
2025-11-25 00:03:22
2
sirothan46
なつ :
共有しすぎてる人ってなんか裏がありそうというか何かを隠すために具体的すぎる予定くれるのかなって思ってる
2025-11-25 14:45:53
2
0326mm2
ひなた :
🥰🥰🥰
2025-12-08 12:06:51
0
su___2731
⌣̈⃝ :
🥰🥰🥰
2025-12-02 13:04:20
0
mikan_nemui
😘 :
遠距離だから会う日決めやすいように共有しようとしてます!
2025-12-13 09:12:12
0
user2080545816037
のりりん :
スマホがない時代なら、こんな事なかったな😭
2026-01-21 05:31:10
0
To see more videos from user @kinumaru0, please go to the Tikwm homepage.

Other Videos

AI for Beginners #12: what is a local model? 💻 A hosted model is a service: send a prompt, pay per token. A local model is a file: download the weights once and run the same model on your own machine. Local requests to Ollama need no API key, and there's no per-token bill. Why the file sizes look odd: quantization. llama.cpp's docs list Llama 3.1 8B at 32.1 GB original size and 4.9 GB at Q4_K_M. Lower precision means a smaller, faster model with some accuracy loss, and you need about that much memory to load it. Where to start - Ollama: install, pull an open model, and it answers on localhost:11434, including an OpenAI-compatible /v1 path. MIT, 186.7k GitHub stars. - llama.cpp: LLM inference in C/C++. MIT, 134.2k stars. GGUF was developed by its creator. - LM Studio: a desktop app. The free plan runs local LLMs with llama.cpp and MLX underneath, plus a local server. - Hugging Face: filter models by GGUF to find files ready to run. What you give up: a quantized 8B model on a laptop is not a frontier model. Go local for privacy or offline work; call a hosted model when the task needs more. AtmoRouter is the cheapest AI API gateway: one OpenAI-compatible endpoint for 100+ models, up to 95% off official price. atmorouter.dev Past episodes: #7 context window, #8 embeddings, #9 hallucination, #10 RAG, #11 fine-tuning. Sources https://docs.ollama.com/api/introduction https://github.com/ggml-org/llama.cpp/blob/master/tools/quantize/README.md https://huggingface.co/docs/hub/en/gguf https://lmstudio.ai/pricing https://github.com/ollama/ollama #AI #LLM #AIforBeginners #LocalLLM #Ollama #llamacpp #OpenSource #AtmoRouter
AI for Beginners #12: what is a local model? 💻 A hosted model is a service: send a prompt, pay per token. A local model is a file: download the weights once and run the same model on your own machine. Local requests to Ollama need no API key, and there's no per-token bill. Why the file sizes look odd: quantization. llama.cpp's docs list Llama 3.1 8B at 32.1 GB original size and 4.9 GB at Q4_K_M. Lower precision means a smaller, faster model with some accuracy loss, and you need about that much memory to load it. Where to start - Ollama: install, pull an open model, and it answers on localhost:11434, including an OpenAI-compatible /v1 path. MIT, 186.7k GitHub stars. - llama.cpp: LLM inference in C/C++. MIT, 134.2k stars. GGUF was developed by its creator. - LM Studio: a desktop app. The free plan runs local LLMs with llama.cpp and MLX underneath, plus a local server. - Hugging Face: filter models by GGUF to find files ready to run. What you give up: a quantized 8B model on a laptop is not a frontier model. Go local for privacy or offline work; call a hosted model when the task needs more. AtmoRouter is the cheapest AI API gateway: one OpenAI-compatible endpoint for 100+ models, up to 95% off official price. atmorouter.dev Past episodes: #7 context window, #8 embeddings, #9 hallucination, #10 RAG, #11 fine-tuning. Sources https://docs.ollama.com/api/introduction https://github.com/ggml-org/llama.cpp/blob/master/tools/quantize/README.md https://huggingface.co/docs/hub/en/gguf https://lmstudio.ai/pricing https://github.com/ollama/ollama #AI #LLM #AIforBeginners #LocalLLM #Ollama #llamacpp #OpenSource #AtmoRouter

About