visionsmemories

LocalLlama is saved!

Posted by danielhanchen@reddit | LocalLLaMA | View on Reddit | 80 comments

Lorem Ipsum

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 2 comments

RTX 5090 Blackwell - Official Price

Posted by Kooky-Somewhere-2883@reddit | LocalLLaMA | View on Reddit | 324 comments

Gambling with language models: One clueless investor's attempt at beating the stock market with ModernBert

Posted by Manwith2plans@reddit | LocalLLaMA | View on Reddit | 23 comments

Llama 3.3 70B drops.

Posted by appakaradi@reddit | LocalLLaMA | View on Reddit | 76 comments

Openai Sora getting leaked was not on my bingo card today wth

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 11 comments

visionsmemories@reddit (OP)

sources [https://huggingface.co/spaces/PR-Puppets/PR-Puppet-Sora](https://huggingface.co/spaces/PR-Puppets/PR-Puppet-Sora) [https://x.com/legit\_rumors/status/1861431113408794898/photo/1](https://x.com/legit_rumors/status/1861431113408794898/photo/1) note that this is sora turbo which means likely not the biggest model they have

Could it be Qwen2.5-Coder 72b 😮??

Posted by notrdm@reddit | LocalLLaMA | View on Reddit | 18 comments

Could it be Qwen2.5-Coder 72b 😮??

Posted by notrdm@reddit | LocalLLaMA | View on Reddit | 18 comments

visionsmemories@reddit

theyre kinda slow at releasing models. for example in fortnite theres a new season every 2 months, if alibaba research were as efficient we would have had qwen 6 by now

Qwen/Qwen2.5-Coder-32B-Instruct · Hugging Face

Posted by Master-Meal-77@reddit | LocalLLaMA | View on Reddit | 162 comments

visionsmemories@reddit

youre correct about their benchmarks being slightly missleading, but cmon man, you get a sota open weights coder model for precisely 0.0$ and the first thing you do is complain? i mean you do you, whatever makes you happy

Qwen/Qwen2.5-Coder-32B-Instruct · Hugging Face

Posted by Master-Meal-77@reddit | LocalLLaMA | View on Reddit | 162 comments

visionsmemories@reddit

your situation is unfortunate https://preview.redd.it/735qqvti9c0e1.png?width=286&format=png&auto=webp&s=be5227dfec0b75475536a3a81ebdf6584069a771 probably just use the 7b q4, or experiment with running 14b or even low quant 32b, though speeds will be quite low due to ram speed bottleneck

Putting together all the AI-powered web search software we know of

Posted by Felladrin@reddit | LocalLLaMA | View on Reddit | 72 comments

visionsmemories@reddit

problem is this seems really good on paper, but as for actual applications - there isnt an immediate advantage you get from it; so you just decide not to make it yourself and so do almost everyone else

Putting together all the AI-powered web search software we know of

Posted by Felladrin@reddit | LocalLLaMA | View on Reddit | 72 comments

Tencent just put out an open-weights 389B MoE model

Posted by girishkumama@reddit | LocalLLaMA | View on Reddit | 183 comments

Tencent just put out an open-weights 389B MoE model

Posted by girishkumama@reddit | LocalLLaMA | View on Reddit | 183 comments

Tencent just put out an open-weights 389B MoE model

Posted by girishkumama@reddit | LocalLLaMA | View on Reddit | 183 comments

Tencent just put out an open-weights 389B MoE model

Posted by girishkumama@reddit | LocalLLaMA | View on Reddit | 183 comments

Tencent just put out an open-weights 389B MoE model

Posted by girishkumama@reddit | LocalLLaMA | View on Reddit | 183 comments

So where’s Qwen2.5-Coder-32B?

Posted by Balance-@reddit | LocalLLaMA | View on Reddit | 27 comments

Exploring AI's inner alternative thoughts when chatting

Posted by Eaklony@reddit | LocalLLaMA | View on Reddit | 52 comments

Benchmark proposal: explain-xkcd

Posted by arnokha@reddit | LocalLLaMA | View on Reddit | 32 comments

Benchmark proposal: explain-xkcd

Posted by arnokha@reddit | LocalLLaMA | View on Reddit | 32 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

visionsmemories@reddit (OP)

screw youtube i wanna see a horror game thats based on training data from idk liveleak. or a racing game made out of ironic tiktoks. or orr the possibilities are wild indeed

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

This is fully ai generated, realtime gameplay. Guys. It's so over isn't it

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 289 comments

visionsmemories@reddit (OP)

https://preview.redd.it/4cqgma6qd6yd1.png?width=1450&format=png&auto=webp&s=7613b3c667d4d6b43bf456c452563195b7f43c8f \> Gameplay is just the start. Soon, most of the internet will be AI-generated i mean, in many ways it aleady is, but i don't think anybody is truly ready for whats already happening source: [xcancel.com/Etched/status/1852089772329869436](http://xcancel.com/Etched/status/1852089772329869436)

So Apple showed this screenshot in their new Macbook Pro commercial

Posted by SandboChang@reddit | LocalLLaMA | View on Reddit | 156 comments

Learning LMs with Journaling

Posted by Mxwhite484@reddit | LocalLLaMA | View on Reddit | 6 comments

3 times this month already?

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 112 comments

3 times this month already?

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 112 comments

3 times this month already?

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 112 comments

3 times this month already?

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 112 comments

visionsmemories@reddit (OP)

source: [https://www.ibm.com/new/ibm-granite-3-0-open-state-of-the-art-enterprise-models](https://www.ibm.com/new/ibm-granite-3-0-open-state-of-the-art-enterprise-models) nobody benchmarks against qwen2.5

Has anybody made a perplexity clone with a higher degree of control?

Posted by ChrisHarles@reddit | LocalLLaMA | View on Reddit | 10 comments

Claude wrote me a script that allows Llama 3.2 1B to simulate Twitch chat

Posted by eposnix@reddit | LocalLLaMA | View on Reddit | 36 comments

Is LLM Studio good?

Posted by Top_Sonic@reddit | LocalLLaMA | View on Reddit | 91 comments

XTC sampler has been merged into llama.cpp mainline

Posted by Master-Meal-77@reddit | LocalLLaMA | View on Reddit | 23 comments

best laptop to run local models for ~$2k

Posted by Sad-Seesaw-3843@reddit | LocalLLaMA | View on Reddit | 13 comments

My First LLM only Build on a Budget. 250€ all together.

Posted by docsnick@reddit | LocalLLaMA | View on Reddit | 34 comments

I finally achieved my AI dream.

Posted by Rombodawg@reddit | LocalLLaMA | View on Reddit | 79 comments

What are prompts and techniques to make LLMs less likely to rephrase what i just wrote, and instead do something more useful than confirming my point of view?

Posted by visionsmemories@reddit | LocalLLaMA | View on Reddit | 8 comments

Hidden Gem: happzy2633/qwen2.5-7b-ins-v3 is an uncensored, CoT finetune with remarkable capabilities

Posted by CryptoSpecialAgent@reddit | LocalLLaMA | View on Reddit | 40 comments

Hidden Gem: happzy2633/qwen2.5-7b-ins-v3 is an uncensored, CoT finetune with remarkable capabilities

Posted by CryptoSpecialAgent@reddit | LocalLLaMA | View on Reddit | 40 comments

What is a good first project to learn how LLM’s work?

Posted by Chimkinsalad@reddit | LocalLLaMA | View on Reddit | 18 comments

LM Studio ships an MLX backend! Run any LLM from the Hugging Face hub on Mac blazingly fast! âš¡

Posted by vaibhavs10@reddit | LocalLLaMA | View on Reddit | 97 comments

LM Studio ships an MLX backend! Run any LLM from the Hugging Face hub on Mac blazingly fast! âš¡

Posted by vaibhavs10@reddit | LocalLLaMA | View on Reddit | 97 comments

Do you agree with this? It has been a two horse race. Google has NEVER been on top in nearly 2 years now. Meta, Mistral and the big Chinese companies are roughly joint 4th but their Open weights business model is based on disrupting the top 3. Amazon and Apple nowhere to be seen.

Posted by Studyw496@reddit | LocalLLaMA | View on Reddit | 169 comments

Open WebUI 0.3.31 adds Claude-like ‘Artifacts’, OpenAI-like Live Code Iteration, and the option to drop full docs in context (instead of chunking / embedding them).

Posted by Porespellar@reddit | LocalLLaMA | View on Reddit | 112 comments