/ais/ - Artificial Intelligence Tools

"In the Future, Entertainment will be Randomly Generated" - some Christian Zucchini

Index Catalog Archive Bottom Refresh
+
-
Name
Options
Subject
Message

Max message length: 12000

Files

Max file size: 32.00 MB

Total max file size: 50.00 MB

Max files: 5

Supported file types: GIF, JPG, PNG, WebM, OGG, and more

E-mail
Password

(used to delete files and posts)
Misc

Remember to follow the Rules

The backup domains are located at 8chan.se and 8chan.cc. TOR access can be found here, or you can access the TOR portal from the clearnet at Redchannit 3.0 (Temporarily Dead).



8chan.moe is a hobby project with no affiliation whatsoever to the administration of any other "8chan" site, past or present.
Use this board to discuss anything about the current and future state of AI and Neural Network based tools, and to creatively express yourself with them. For more technical questions, also consider visiting our sister board about Technology
(134.07 KB 1024x1024 lmg_.jpg)

/lmg/ - local models general Anonymous 04/16/2025 (Wed) 06:15:26 No. 6258
/lmg/ - a general dedicated to the discussion and development of local language models. ►News >(04/14) GLM-4-0414 and GLM-Z1 released: https://hf.co/collections/THUDM/glm-4-0414-67f3cbcb34dd9d252707cb2e >(04/14) Nemotron-H hybrid models released: https://hf.co/collections/nvidia/nemotron-h-67fd3d7ca332cdf1eb5a24bb >(04/10) Ultra long context Llama-3.1-8B: https://hf.co/collections/nvidia/ultralong-67c773cfe53a9a518841fbbe >(04/10) HoloPart: Generative 3D Part Amodal Segmentation: https://vast-ai-research.github.io/HoloPart ►News Archive: https://rentry.org/lmg-news-archive ►Glossary: https://rentry.org/lmg-glossary ►Links: https://rentry.org/LocalModelsLinks ►Official /lmg/ card: https://files.catbox.moe/cbclyf.png ►Getting Started https://rentry.org/lmg-lazy-getting-started-guide https://rentry.org/lmg-build-guides https://rentry.org/IsolatedLinuxWebService https://rentry.org/tldrhowtoquant ►Further Learning https://rentry.org/machine-learning-roadmap https://rentry.org/llm-training https://rentry.org/LocalModelsPapers ►Benchmarks LiveBench: https://livebench.ai Programming: https://livecodebench.github.io/leaderboard.html Code Editing: https://aider.chat/docs/leaderboards Context Length: https://github.com/hsiehjackson/RULER Japanese: https://hf.co/datasets/lmg-anon/vntl-leaderboard Censorbench: https://codeberg.org/jts2323/censorbench GPUs: https://github.com/XiongjieDai/GPU-Benchmarks-on-LLM-Inference ►Tools Alpha Calculator: https://desmos.com/calculator/ffngla98yc GGUF VRAM Calculator: https://hf.co/spaces/NyxKrage/LLM-Model-VRAM-Calculator Sampler Visualizer: https://artefact2.github.io/llm-sampling ►Text Gen. UI, Inference Engines https://github.com/lmg-anon/mikupad https://github.com/oobabooga/text-generation-webui https://github.com/LostRuins/koboldcpp https://github.com/ggerganov/llama.cpp https://github.com/theroyallab/tabbyAPI https://github.com/vllm-project/vllm
Anonymous 04/20/2025 (Sun) 04:13:41 Id: b1a103 No. 7968
>>7358 >Implying that 50% of /lmg/ discussion wasn't always about trying out whatever new meme finetune
Anonymous 04/20/2025 (Sun) 04:30:37 Id: 91dfd1 No. 7974
>>7962 I'll give it a try next time ty
Anonymous 04/20/2025 (Sun) 04:31:50 Id: 0e20ba No. 7975
(341.04 KB 1920x5224 retarded.webp)

(44.27 KB 1734x302 retardedtwice.webp)

>>7965 i love ids
Anonymous 04/20/2025 (Sun) 04:53:56 Id: 1037e2 No. 7980
>>7975 Based. Easy to get around though. I post through a vpn and get a new ID every time without changing anything. Not intentional, I like IDs.
Anonymous 04/20/2025 (Sun) 06:45:30 Id: dfb40b No. 7999
>riverwind Is this a trolling model? I keep getting shilled by products.
petr 04/20/2025 (Sun) 06:46:13 Id: 4c17cd No. 8001
>>7999 kek
Anonymous 04/20/2025 (Sun) 06:46:30 Id: 4c17cd No. 8002
>>8001 >001 AAAAAAAAAACCCCCCCCKKKK
Anonymous 04/20/2025 (Sun) 07:23:31 Id: 8b4562 No. 8012
>>7999 yes its a troll model, unironically great at what its made to do
Anonymous 04/20/2025 (Sun) 07:27:52 Id: 5fa221 No. 8016
>>7999 pretty sure it was an april fools day project that wasnt ready in time
Anonymous 04/20/2025 (Sun) 07:59:25 Id: 48eb0b No. 8021
>>7959 Probably, the guy who makes most of the posts there replied to himself here twice >>7237 >>7264 >>7275 >>7276
Anonymous 04/20/2025 (Sun) 08:23:14 Id: f38237 No. 8027
>>8021 What the fuck lmao what a weird cunt. if 4chan ever comes back IDs need to be on every board to out freakshows like this
Anonymous 04/20/2025 (Sun) 08:31:29 Id: 8b4562 No. 8028
>>8021 Why the fuck are You giving him (You)s
Anonymous 04/20/2025 (Sun) 08:33:34 Id: 0e20ba No. 8029
(248.28 KB 828x938 1726959941008009.jpg)

>>8021 What causes one to behave this way?
Anonymous 04/20/2025 (Sun) 08:39:24 Id: 2f83dc No. 8031
>>8028 you's are not currency dont be a faggot, this person deserves to be pointed out and shamed
Anonymous 04/20/2025 (Sun) 08:41:43 Id: 8af965 No. 8032
>>8029 Mental illness.
Anonymous 04/20/2025 (Sun) 08:43:13 Id: 8b4562 No. 8033
(313.07 KB 3360x2640 a48ecf3009ef2c35f123aa29b5646be96fb6d7cb87dc39d62ad174b0413a08ae.png)

https://github.com/JohannesGaessler/elo_hellm >Elo HeLLM is a project for establishing a ranking based on Elo ratings between large language models. The context is that I'm working on training code for llama.cpp. llama.cpp has methods for estimating the quality loss from quantization but it lacks methods for estimating the quality of a model in absolute terms or for making comparisons between different models. I intend to co-develop this project with the llama.cpp training code for quality control. The approach is to merge an arbitrary number of quality metrics into a single Elo rating for a model using statistical methods. One category of such quality metrics are simply the results of language model benchmarks such as MMLU. Results from competitive games such as Chess can also be used (not yet implemented).
Anonymous 04/20/2025 (Sun) 08:55:27 Id: bb0219 No. 8036
Hey wait wtf. I just noticed that my post here >>7237 has the same ID as a bunch of other posts in the thread that aren't mine. I'm serious. Also, I don't see "(You)" in the replies. I'm getting spooked what the hell.
Anonymous 04/20/2025 (Sun) 08:56:25 Id: bb0219 No. 8037
>>8036 Why is my id different ahhhhhhh.
Anonymous 04/20/2025 (Sun) 08:57:23 Id: 8b4562 No. 8038
(196.71 KB 269x375 1734017971365721.gif)

>tfw anons that leave /lmg/ for too long get assimilated into petra after all
Anonymous 04/20/2025 (Sun) 08:57:30 Id: 1bac8e No. 8039
>>8036 fake until proven gay
Anonymous 04/20/2025 (Sun) 08:58:57 Id: 0e20ba No. 8040
>>8038 >petrified petra is a gorgon
Anonymous 04/20/2025 (Sun) 09:00:17 Id: bb0219 No. 8041
>>8038 >>8039 But seriously though this is creepy. Are the mods messing with me? Did I get hacked? How am I even supposed to get proof in this situation?
Anonymous 04/20/2025 (Sun) 09:02:25 Id: 48eb0b No. 8043
>>8041 Why do you care? Even if you are telling the truth you are anonymous and have no identity worth protecting.
Anonymous 04/20/2025 (Sun) 09:06:39 Id: 8af965 No. 8044
>>8041 Your IP could have changed and some other guy has gotten your exact previous one. Which is probably less likely than winning the lottery.
Anonymous 04/20/2025 (Sun) 09:06:48 Id: 8b4562 No. 8045
serial expetriments lain
Anonymous 04/20/2025 (Sun) 09:06:54 Id: b1e463 No. 8046
>>8041 In all probability, someone is just using the same VPN.
Anonymous 04/20/2025 (Sun) 09:07:49 Id: 0e20ba No. 8047
>>8046 this is true. watch me change me id by changing my vpn
Anonymous 04/20/2025 (Sun) 09:08:18 Id: 6f10fa No. 8048
>>8047 Sex with AI.
Anonymous 04/20/2025 (Sun) 09:08:34 Id: 465644 No. 8049
>>8048 as shrimple as that
Anonymous 04/20/2025 (Sun) 09:08:35 Id: 8b4562 No. 8050
anons what if he hacked 8chan too?
Anonymous 04/20/2025 (Sun) 09:10:27 Id: 8af965 No. 8051
>>8050 He won't get away with it on 16chan.
Anonymous 04/20/2025 (Sun) 09:11:28 Id: 8b4562 No. 8052
Christ is risen Hitler's birthday Kikes seething
Anonymous 04/20/2025 (Sun) 09:13:03 Id: bb0219 No. 8053
>>8043 Why would I not care? ID's serve a purpose, and people are treating them as something that has a purpose, so if they can be undermined, then we can't really treat them the same anymore. And I don't see why it someone wouldn't be concerned if they were the target of some mod trolling or other activity, assuming this wasn't due to a bug or some one in a million chance. >>8044 Last I checked I have a static IP. I do use librewolf though which might change my canvas/fingerprint around sometimes, does this site use other indicators other than IP to assign an ID? If so then perhaps that's why. >>8046 I wasn't using a VPN when I made that first post, and I'm not using one right now. I did use a VPN to take a look at gay 4chan tho.
Anonymous 04/20/2025 (Sun) 09:45:12 Id: 7d7a2a No. 8067
I got different IDs too even though I have (supposedly) static IP.
Anonymous 04/20/2025 (Sun) 09:53:11 Id: 65ba16 No. 8068
>be me >rode the wave of AI cooming before proxies dried up and became hoarded by people >forget about AI cooming for a bit >get a 7900XTX for vidyagames >only now i realize i could run a model locally and coom my brains out Ok, I've got Ooba set up, what NSFW models would you suggest for 24 GB of VRAM and 32 GB of RAM?
Anonymous 04/20/2025 (Sun) 09:55:50 Id: 8b4562 No. 8069
>>8068 nevoria 70b or whatever its called
Index Catalog Archive Top Reply
67
Forms
Delete
Report
Quick Reply