Haha, that makes way more sense — I was worried I was behind the curve, I’m just a beginner, which is good! Better than being behind!
Joseph Jones PRO
KlondikeDev
AI & ML interests
Restoring the Digital Millennium. https://kunix.org/digital-millennium.html
Recent Activity
View all activity
Organizations
Only a week?? I’ve been doing it for a long time and some of your stuff feels over my head!
Of course, that’s completely understandable. I’m honestly the same way — which I guess seems strange considering I offered it. I just really like what you do — and I think you could make something really impressive with more hardware. You clearly have the skill.
Yeah, that’s fair enough. If you want, I’m setting up a NVIDIA Tesla V100, and I have a 3060 in the same machine — I’d be willing to test it, no strings attached if you wanted to — you could just tell me the datasets.
Oh shoot, I just read that you had done continued training. That’s my fault for not reading that. Ha, I just read the whole model card — that’s a big part of it. I should have read it before guessing. But, you should still check with the datasets, I think.
Have you tried continued pretraining? That’s what I did with Boris-1.3, and it helped dramatically. Try running some on one dataset, then another — see what benchmarks each one moves and do that accordingly. That’s what I did. If you do WSD LR, you can continue from stable and it doesn’t jack it up terribly.
You could always try Gated DeltaNet2? Thats part of what I’m trying with Boris-2 — it reaches near parity or surpasses Transformers at a small scale, just not recall.
It might also just be reaching the kinda parameter limit, you might have squeezed the most you can out of those few parameters.
I could be completely wrong, I’m just starting out.
posted an update about 4 hours ago
Post
18
The Small Language Model Consortium has reached 10 members!
Anybody is welcome to join, via two methods:
1. Starting a community discussion on the group’s README
2. Receiving an invite from me: @KlondikeDev
The Discord Community is also live! Feel free to join, even if you aren’t in the consortium, to discuss Small Language Models, or even just AI in general!
https://discord.gg/FngBKjja4
You’re more than welcome to apply!
— KlondikeDev
Anybody is welcome to join, via two methods:
1. Starting a community discussion on the group’s README
2. Receiving an invite from me: @KlondikeDev
The Discord Community is also live! Feel free to join, even if you aren’t in the consortium, to discuss Small Language Models, or even just AI in general!
https://discord.gg/FngBKjja4
You’re more than welcome to apply!
— KlondikeDev
reacted to RiverRider's post with 🔥 about 5 hours ago
Post
166
Two rival models, one memory.
Four multimodal models from four different companies (two at a time). One of them indexes a thousand photographs. Then a different company's model searches that index, having never been trained to work with it.
The same pictures come back.
Cross-company retrieval is statistically indistinguishable from a model searching its own index: retention 0.988, 95% CI [0.955, 1.023], across 12 cross directions and 1,000 held-out items. The interval contains 1.0, so we report indistinguishability and keep the interval attached to the number. Four of the twelve cross directions score above the within-vendor baseline of the gallery they search.
A single linear tower also places image, audio and video beside text in one searchable space, beating one tower per modality on every modality. Video gains most, 0.076.
Two strips of results for the same request, one per company:
RiverRider/srt-omni-demo
The four hosts:
Qwen/Qwen3-Omni-30B-A3B-Instruct
google/gemma-4-31B-it
mistralai/Mistral-Small-3.1-24B-Instruct-2503
rhymes-ai/Aria
States, fitted towers and the manifest are published, so you can add a host we have not tried:
RiverRider/srt-omni-xvendor-towers
Scope. The four-vendor result covers images, since two of the hosts handle images only. Raw cosine between unrelated items on these states is +0.869 and raw retrieval sits at chance, so centering is not optional. All four hosts train on overlapping web-scale corpora, so this shows the readable structure survives a change of vendor, architecture and training run. The stronger reading, that independently trained systems would converge on that structure, remains untested and we do not assert it.
Four multimodal models from four different companies (two at a time). One of them indexes a thousand photographs. Then a different company's model searches that index, having never been trained to work with it.
The same pictures come back.
Cross-company retrieval is statistically indistinguishable from a model searching its own index: retention 0.988, 95% CI [0.955, 1.023], across 12 cross directions and 1,000 held-out items. The interval contains 1.0, so we report indistinguishability and keep the interval attached to the number. Four of the twelve cross directions score above the within-vendor baseline of the gallery they search.
A single linear tower also places image, audio and video beside text in one searchable space, beating one tower per modality on every modality. Video gains most, 0.076.
Two strips of results for the same request, one per company:
RiverRider/srt-omni-demo
The four hosts:
Qwen/Qwen3-Omni-30B-A3B-Instruct
google/gemma-4-31B-it
mistralai/Mistral-Small-3.1-24B-Instruct-2503
rhymes-ai/Aria
States, fitted towers and the manifest are published, so you can add a host we have not tried:
RiverRider/srt-omni-xvendor-towers
Scope. The four-vendor result covers images, since two of the hosts handle images only. Raw cosine between unrelated items on these states is +0.869 and raw retrieval sits at chance, so centering is not optional. All four hosts train on overlapping web-scale corpora, so this shows the readable structure survives a change of vendor, architecture and training run. The stronger reading, that independently trained systems would converge on that structure, remains untested and we do not assert it.
reacted to CodeSoft's post with ❤️ about 21 hours ago
Post
1401
Wow, SLM Arena is getting a lot of traffic! Thank you guys for showing your interest!
To handle the growing demand, I’m moving SLM Arena from a CPU Space to a ZeroGPU Space. Hopefully, this will let me add more models to SLM Arena while keeping it running fast.
I've also added a separate arena + leaderboard for base models!
If there are any models or features you’d like to see, let me know in a reply to this post or in a Community post on the Space!
To handle the growing demand, I’m moving SLM Arena from a CPU Space to a ZeroGPU Space. Hopefully, this will let me add more models to SLM Arena while keeping it running fast.
I've also added a separate arena + leaderboard for base models!
If there are any models or features you’d like to see, let me know in a reply to this post or in a Community post on the Space!
- 17 replies
replied to their post 1 day ago
I invited you, did it not show up? You’re more than welcome in!
replied to their post 2 days ago
I’ve never merged anything and I don’t have an api wrapper😭😭 bro go away💔
replied to their post 2 days ago
Too true dude like they could at least be original
replied to their post 2 days ago
Son I’m crine😭😭 worlds most ai generated message ever
You’re not even arguing for anything, not for me, nor Dan — and you’re a bot account💔 son
replied to their post 2 days ago
Alright, no problem! Sorry for the late reply!
replied to their post 2 days ago
Of course! I’m no expert either! Would you like an invite?
replied to their post 3 days ago
Here, can you join the discord? We’re all active in there, and it would be way faster to talk in there
replied to Banaxi-Tech's post 3 days ago
Awesome! BananaMindOS is such a cool project
reacted to Banaxi-Tech's post with ❤️ 3 days ago
Post
2843
We’re excited to release BananaMind OS 2.0, a major update to our portable operating system for running AI models locally.
BananaMind OS runs directly from an ISO without Linux, installation, a cloud connection, or modifying your disks.
The new Version 2.0 adds a graphical interface with mouse support, a model library, multi-turn chat, configurable KV cache, maximum tokens and temperature
controls, automatic x87/SSE/SSE2 CPU detection, BIOS support, native UEFI support and a dedicated 486 compatibility mode.
Model weights are not loaded during GRUB or startup. Only the model catalog is read, and the selected model is loaded after BananaMind OS starts.
We now have a new .litemodel format supports multiple architectures, including BananaMind models, SmolLM, SmolLM2, GPT-X2.5, min-spark 1.1 and Rose-Mini.
Check it out at:
https://github.com/BananaMind/BananaMindOS
You can build an ISO yourself using the .sh or .bat script. Select the models and quantizations you want, and the builder will download, quantize
and package them locally.
Prebuilt 10MB, 25MB, 100MB and 250MB model presets are available here:
https://github.com/BananaMind/BananaMindOS/releases/tag/v2.0.0
Use the regular preset ISOs for BIOS and GRUB, including the 486 compatibility mode.
Warning: Im currently uploading the ISOs, all up to 100MB is present, 250MB is getting uploaded
Use the files ending in -uefi.iso for the native x86-64 UEFI graphical frontend and improved firmware mouse support.
The preset name describes the RAM class of the individual included models. Models remain on the ISO until selected, so including multiple models does not load all of them into RAM.
486DX with an x87 FPU, Pentium and newer x86 processors are supported. Modern x86-64 computers are supported through UEFI. It currently doesent support processors without an FPU.
Comment if you want me to run it on 0.04MHz (pls dont)
Video Credit:
Song: Matzan - Redesigned
Music provided by NoCopyrightSounds
BananaMind OS runs directly from an ISO without Linux, installation, a cloud connection, or modifying your disks.
The new Version 2.0 adds a graphical interface with mouse support, a model library, multi-turn chat, configurable KV cache, maximum tokens and temperature
controls, automatic x87/SSE/SSE2 CPU detection, BIOS support, native UEFI support and a dedicated 486 compatibility mode.
Model weights are not loaded during GRUB or startup. Only the model catalog is read, and the selected model is loaded after BananaMind OS starts.
We now have a new .litemodel format supports multiple architectures, including BananaMind models, SmolLM, SmolLM2, GPT-X2.5, min-spark 1.1 and Rose-Mini.
Check it out at:
https://github.com/BananaMind/BananaMindOS
You can build an ISO yourself using the .sh or .bat script. Select the models and quantizations you want, and the builder will download, quantize
and package them locally.
Prebuilt 10MB, 25MB, 100MB and 250MB model presets are available here:
https://github.com/BananaMind/BananaMindOS/releases/tag/v2.0.0
Use the regular preset ISOs for BIOS and GRUB, including the 486 compatibility mode.
Warning: Im currently uploading the ISOs, all up to 100MB is present, 250MB is getting uploaded
Use the files ending in -uefi.iso for the native x86-64 UEFI graphical frontend and improved firmware mouse support.
The preset name describes the RAM class of the individual included models. Models remain on the ISO until selected, so including multiple models does not load all of them into RAM.
486DX with an x87 FPU, Pentium and newer x86 processors are supported. Modern x86-64 computers are supported through UEFI. It currently doesent support processors without an FPU.
Comment if you want me to run it on 0.04MHz (pls dont)
Video Credit:
Song: Matzan - Redesigned
Music provided by NoCopyrightSounds
- 3 replies
replied to their post 3 days ago
And there already is a way — by waiting, like he said was fine.
It will be like a couple more hours dude, relax.