anthony_s_225
Region: US
Tuesday 09 December 2025 04:51:01 GMT
506185
47283
520
2001
Music
Download
Comments
xxx :
ollama + langgraph + neovim
2025-12-11 03:00:00
170
ricarnffjnu :
Codex or Claude code both mog your LLM btw.
2025-12-09 05:53:52
2121
CatMan :
Someone discovered ollama today, but if that model is accurate, that’s pretty cool.
2025-12-10 02:26:03
1052
Leo San :
literally why, just download chinese ones lol.
2025-12-09 11:15:59
1955
user5964826091044 :
this is really sick that visualization is dope
2025-12-09 16:04:41
1657
𝓚𝓪𝔀𝓪𝓲𝓲 :
Neural Network vs Neuron cell
2025-12-12 06:15:35
315
Him :
Why use a local worse performing llm when you can use a public llm that performs better. Unless of course you’re fine tuning it for a specific task or the extra privacy.
2025-12-19 02:49:01
38
Og Gaming :
How do you create your own LLM
2025-12-09 20:21:25
185
alec :
I really like the visualization, could you expand on what you’re displaying exactly? I would assume it’s not actual neuron connections but something like a RAG knowledge graph? Very cool stuff!
2025-12-09 17:57:00
60
Ghol29 :
Can you ask it illegal questions?
2025-12-10 00:37:15
52
Momen Walied :
Any one want to run local llm (not prefer this) but still you can download Ollama and download any model version like Gemma3 based on your GPU VRAM. If you want use vllm you can download any models you need from huggingface and run it but that is way make model more big because it working with fp16
2025-12-11 15:39:56
14
doe :
have a good laptop. download llm studio. install uncensored chat gpt. run it offline.
2026-01-02 05:15:15
29
Ace.. :
Making my own slm from scratch no api nor hugging face
2025-12-24 20:43:26
7
JackCooper :
how the fuck did you get that graphic that's dope
2026-01-09 20:22:57
34
️ :
i have a “local” llm but it’s just a mistral model running on my gpu
2025-12-09 12:21:11
125
C r u x u h :
Apple's M Series architecture is leading the way 🔥
2025-12-09 16:51:15
26
yessitsJay :
it’s not worth it yet, i get the appeal and am super excited for when they’re more efficient to run
2025-12-10 04:32:44
13
Oxygean :
Why tho... Deepseek is completely opensource just use it locally
2025-12-22 06:03:49
10
julian :
What’s the point of having ur own when it’s still js downloaded like ollama and u can js use online ai which has way more params?
2025-12-09 15:05:17
66
balluptop :
whats that visual? is that actually a representation of your llms neural network processing your inputs in real time? thats sick. is it opensource?
2025-12-09 06:18:48
56
𝔔𝔡𝔫𝔵 :
what tool are you using for the graph
2025-12-10 15:40:25
5
wizfur :
How did you learn about it? I’ve been thinking of self hosting one
2025-12-09 07:40:40
12
[ondyhop@verity~]$ :
what is a LLM?
2025-12-10 12:23:45
18
GuiguiLabuse :
mine is in my own cognitive system because of the real quantum particle we created from a Qubit. 🤷😎🤣
2025-12-30 21:23:24
1
Gaspard V :
Guys I’m making a new 300m param llm for fun lol this is version 5 I’ve done many smaller ones, so my pretraining is only 70k steps batch 16 seqlen 256 and instruct I wanna do like 27k steps batch 16 seqlen 256 but idk if that’s too much instruction training or too little…. Also I have limited gpu time so I can’t go more steps for this version please help
2026-05-01 23:58:50
0
To see more videos from user @anthony_s_225, please go to the Tikwm
homepage.
