NLP. NMT. Main author of Marian NMT. Research Scientist at Microsoft Translator. This account formerly used the handle @marian_nmt which is now parked.
Cascadia
Joined April 2017
- Tweets5.8K
- Following401
- Followers2.5K
- Likes4.1K
Pinned Tweet
Blog post for NVIDIA DGX Spark havers:
Linux is hiding 4.1 GB of RAM from your DGX Spark.
On my 2× Spark setup, getting it back took my GLM-5.3-Flash KV pool from 262k → 938k tokens.
Two tricks I haven’t seen discussed together:
• 64 KiB kernel
• reclaim the headless display carveout
3
6
604
I wonder if GPT-5.6 Sol actually benefits from hyping itself up in its thinking traces:
"There’s a lot here to manage, but I’ll stay focused!"
"This is quite complex, but I can do it!"
Would suppressing this kind of self-encouragement hurt its performance?
2
3
477
Probably the fault of this guy who is running Kimi K3 on 16 OEM Sparks.
reddit.com/r/LocalLLaMA/comm…
1
179
I don't like the silence of the GPT 6.x generation of models while they work. GPT 5.6 Sol's communication style is pretty much ideal. I wonder if there will be work to bring that back.
2
9
497
I keep being amazed that you can just do things now, and the only blocker is "actually thinking of them".
For years, my old Surface Book 3 didn’t have camera driver on Linux. Today, I told Copilot to fix it, and two hours later I’m testing my camera in WhatsApp.
github.com/emjotde/surface-b…
I should probably contribute this somehow.
1
4
476
• Mapped both sensors, CSI-2 routing and RAW10 formats.
• Traced front-camera failures to receiver reset timing—and fixed it.
• Ported IPU4P to Linux 6.8 without disabling Secure Boot.
• Turned raw frames into upright color video.
• Added on-demand capture: cameras stay listed, but sleep when unused.
• Fixed GNOME Camera discovery, crashes and switching.
• Published the working source, patches and setup guide.
1
156
PSA no. 2: Turns out that is an SGLang serving bug (github.com/sgl-project/sglan…). GLM and MiniMax were hit by the same bug. The tool-storming of Flash seems to be fixed with the MOPD version.
![[Bug] [Responses API] Replayed Codex turn is split into three assistant blocks (phase check in...](https://pbs.twimg.com/card_img/2119371571634765824/peclOGkX?format=jpg&name=800x419)
PSA: seems like both MiMo V2.6 Flash variants seem basically impossible to get working reliably as coding agents.
The original loops/tool-storms, MOPD randomly stops or drops out, and neither has made it through a single SWE-bench task cleanly for me.
User error not excluded.
1
543
I have been looking into soldering more RAM onto the 128GB versions, but I expect now that jerry-rigging the 64GB versions back to 128GB might become a whole new industry. If the firmware/BIOS is the same, the chips demonstrably exist.
2
255
Spark DGX 128GB price hike to 7K USD!
marketplace.nvidia.com/en-us…
Individual compute inequality, I realized recently, might be one of the bigger societal inequalities we are seeing right now.
Having the perk of near-unlimited frontier model access for reasonable private purposes must be one of the biggest benefits that working for big tech offers right now.
Full write-up, step-by-step instructions, and scripts:
junczys.net/blog/recovering-…
1
4
151
Or the no-frills GitHub version:
github.com/emjotde/spark-ram…
2
70