WayneSaintGermane

We go back

[WayneSaintGermane @W_SaintGermane]

@Alezander907 @browser_use Thanks for the response. You’re right, I didn’t see that, but what I did see was your team retweeting the successes while ignoring the DMs. Busy day, I get it, but a post to acknowledge the users stuck in purgatory would have instantly relieved some of the scammy vibes.

[WayneSaintGermane @W_SaintGermane]

@Alezander907 @browser_use I appreciate your efforts and the learning experience this presented for your team. People gave you their addresses after being offered a gift from a trusted service, only to be denied and told to “prompt harder”. Lessons learned all-around.

[WayneSaintGermane @W_SaintGermane]

@AlexanderM72870 @TrevorCampbell_ @AlexanderM72870 - I think our sessions are borked beyond repair and the model won’t even try anymore. I’m guessing that new sessions starting after the banking issue resolved are going through.

[WayneSaintGermane @W_SaintGermane]

@sawarin41 @browser_use Exact same experience here, along with a fair amount of gaslighting (accusing me of fabricating screenshots to steer its offer) after it forgot what it offered on two occasions during the conversation. @browser_use is a fantastic tool, but this has been an unfortunate experience.

[WayneSaintGermane @W_SaintGermane]

@browser_use - Love the effort, but it forgot it offered me a Raspberry Pi twice and accused me of lying, then saw the price and rescinded the offer. Talked me down to a flash drive, then the banked flagged the charge and declined it.

[WayneSaintGermane @W_SaintGermane]

Learning LLMs in the open with @browser_use — trials, errors and all.

[WayneSaintGermane @W_SaintGermane]

29 days until GPT-OSS’ first birthday. @OpenAI how are we celebrating? 🚢?

[WayneSaintGermane @W_SaintGermane]

Global superpowers interfering with the most significant technology in human history wasn’t unexpected, but my vision of “Artificial General Intelligence benefitting all of humanity” has clouded.

[WayneSaintGermane @W_SaintGermane]

Absolutely a skill issue. Cut down the model_context_window and it started working. Leaving this up for lessons learned in case anyone gets stuck.

[WayneSaintGermane @W_SaintGermane]

More than likely a skill issue, but is auto-compaction just not a thing in Codex when using a local model? I’ve got model_auto_compact_token_limit set at 60% and it just doesn’t kick in.

[WayneSaintGermane @W_SaintGermane]

All conferences should have a virtual option. I don’t wanna drive, fly, network, drink, sleep in a different bed, or walk 10 miles to hear a half-assed talk. That’s cool if your company lets you get a paid vacation, but I got other stuff going on, man.

[WayneSaintGermane @W_SaintGermane]

@ChrissGPT Homeboy’s built like a couch and can’t get all the clothes in the basket. I’d be spending $450/month to follow it around worried all the time.

[WayneSaintGermane @W_SaintGermane]

@apples_jimmy Perfect time to release GPT-OSS 2 @sama

[WayneSaintGermane @W_SaintGermane]

Why do we need a Codex-branded macro keyboard? I’m not here to yuck anyone’s yum, but if you’re a shortcut-junkie, keypads are like $10. I want OpenAI hardware to change the world. Maybe I’ll be wrong, but this ain’t it.

[WayneSaintGermane @W_SaintGermane]

A little embarrassed to admit I still have all three discretionary Codex resets banked.

[WayneSaintGermane @W_SaintGermane]

@LottoLabs Sol = Venti / Terra = Grande / Luna = Tall

[WayneSaintGermane @W_SaintGermane]

Can’t wait until memory production increases so hardware prices can come back down.

Wait… what was that?

They’re not coming back down?

[WayneSaintGermane @W_SaintGermane]

@AndrewCurran_ https://t.co/k71GrWbpuB

[WayneSaintGermane @W_SaintGermane]

• US blocks foreign (and domestic) access to frontier models due to security risks.
• Domestic access restored with ID verification (tracking).
• China overtakes the US in AI capability.
• US blocks domestic access to foreign models due to security risks.

Is that the plan?