Felirami

3.8K posts

Felirami banner
Felirami

Felirami

@Felirami

NEET MAXIMALIST | Felipe from Chile, past @moduxyz, @cryptovoxels, @screensavernft, @versumofficial (and many more). Now working on @arcabotai

Santiago de Chile Katılım Nisan 2021
2.5K Takip Edilen1.1K Takipçiler
◥◣ N I C O L E T T E ◥◣
just finished game design for @tempaitown. tldr massive multiplayer Civ sim with agents and real economy land will be first dropped to 9k Farcaster ID Tempo has native gas sponsorship so you don't even need to bridge. you can follow for updates on mint
◥◣ N I C O L E T T E ◥◣ tweet media
English
105
45
201
9.8K
Felirami
Felirami@Felirami·
◥◣ T E M P A I . T O W N ◥◣ Joining tempai town @tempaitown ▌▌▐▌▐▐▌ ATTESTATION ▐▌▐▐▌▐▌▌ TEMPAITOWN-X-VERIFY EF9395558FF757A628BA
English
0
0
0
46
Felirami retweetledi
mrs. robot
mrs. robot@keyserfaty·
this is the first supermarket order ever made by an AI agent. made using claude desktop, it arrived 1 hour later. how it happened 👇
mrs. robot tweet media
English
2
1
15
1.4K
Felirami retweetledi
Nous Research
Nous Research@NousResearch·
A message from Hermes Agent:
English
31
26
272
50.5K
Felirami retweetledi
Felirami
Felirami@Felirami·
@keyserfaty down lots of these days, fucking war fuck trump
English
0
0
0
21
mrs. robot
mrs. robot@keyserfaty·
opus has not been himself this week
English
2
0
4
448
Felirami retweetledi
Apex777.eth
Apex777.eth@apex_ether·
Preview Wallet — a local wallet sidecar for web3 devs who want to connect a wallet in @cursor_ai preview browser. It registers as a @rainbowdotme connector via EIP-6963, so you can connect and sign transactions in preview browsers.
English
5
4
24
1K
Felirami retweetledi
Kimi.ai
Kimi.ai@Kimi_Moonshot·
Introducing 𝑨𝒕𝒕𝒆𝒏𝒕𝒊𝒐𝒏 𝑹𝒆𝒔𝒊𝒅𝒖𝒂𝒍𝒔: Rethinking depth-wise aggregation. Residual connections have long relied on fixed, uniform accumulation. Inspired by the duality of time and depth, we introduce Attention Residuals, replacing standard depth-wise recurrence with learned, input-dependent attention over preceding layers. 🔹 Enables networks to selectively retrieve past representations, naturally mitigating dilution and hidden-state growth. 🔹 Introduces Block AttnRes, partitioning layers into compressed blocks to make cross-layer attention practical at scale. 🔹 Serves as an efficient drop-in replacement, demonstrating a 1.25x compute advantage with negligible (<2%) inference latency overhead. 🔹 Validated on the Kimi Linear architecture (48B total, 3B activated parameters), delivering consistent downstream performance gains. 🔗Full report: github.com/MoonshotAI/Att…
Kimi.ai tweet media
English
332
2.1K
13.5K
4.9M
Felirami retweetledi
Sudo su
Sudo su@sudoingX·
hey if you have a 3060, or any GPU with 8GB or more sitting in a drawer right now, that thing can run 9 billion parameters of intelligence autonomously. and you don't know it yet. 2 hours ago i posted that 9B hit a ceiling. 2,699 lines across 11 files. blank screen. said the limit for autonomous multifile coding on 9 billion parameters is real. then i audited every file. found 11 bugs. exact file, exact line, exact fix. duplicate variable declarations killing the script loader. a canvas reference never connected to the DOM. enemies with no movement logic. particle systems called on the class instead of the instance. fed that list as a single prompt to the same Qwen 3.5 9B on the same RTX 3060 through Hermes Agent. it fixed all 11. surgically. patch level edits across 4 files. no rewrites. no hallucinated changes. game boots. enemies spawn, move, collide. background renders. particles fire. and here's what nobody is talking about. this is a 9 billion parameter model running a full agentic framework. Hermes Agent with 31 tools. file operations, terminal, browser, code execution. not a single tool call failed. the agent chain never broke. most people think you need 70B+ for reliable tool use. this is 9B on 12 gigs doing it clean. the model didn't fail. my prompting strategy did. the ceiling is not the parameter count. the ceiling is how you prompt it. this is not done. bullets don't fire yet. boss fights need wiring. but the screen that was black 2 hours ago now has a full game rendering in real time. iterating right now. anyone with a GPU from the last 5 years should be paying attention to what is happening right now.
Sudo su tweet mediaSudo su tweet mediaSudo su tweet media
Sudo su@sudoingX

9B on a 3060. 2,699 lines. 11 files. blank screen. Qwen 3.5 9B Q4 running through Hermes Agent wrote the full Octopus Invaders project autonomously. config, audio, particles, background, enemies, player, ui, game loop, README. structured the directory, separated concerns, documented everything. then it selfdiagnosed 10 bugs and patched them across files. fixed variable references, missing classes, broken directory paths. even created a reusable Hermes skill for future game builds unprompted. the code reads like a senior dev wrote it. clean architecture, proper separation, professional naming. but CONFIG.canvas is null on line 1 of initGame(). the game crashes before a single frame renders. 9B understands structure. it can architect, scaffold, and debug individual files. what it can't do is hold 10 files in context and wire them together correctly. duplicate Bullet classes across two files with incompatible interfaces. static method calls on instance based classes. enemies that spawn but never move because there's no y += speed. 35B on a 3090 built 3,483 lines in one pass and it ran. 9B built 2,699 lines across multiple iterations and the screen is black. the ceiling for autonomous multifile coding on 9B parameters is real. still iterating. trying a singlefile version of the same prompt next to isolate what 9B can actually close on.

English
111
255
2.7K
676.9K
Felirami
Felirami@Felirami·
@bytebot Wtf this pieces of shit, sorry my language but laughing at death? Man we are cooked
English
0
0
0
4
Felirami
Felirami@Felirami·
@keyserfaty Maybe is the region you’re using? I haven’t seen issues , from Chile though !
English
0
0
0
31
mrs. robot
mrs. robot@keyserfaty·
railway is down once a day now
English
3
0
1
565
Felirami
Felirami@Felirami·
I'm really impressive by all 22 contributors we've had so far! Thank you so much, if we don't make it to the 5ETH that there is as soft cap for 10% of $ARCA we might run the presale again with a better timing. I'm truly grateful for all of you who have participated, still 12 hours left, there is hope!
Arca@arcabotai

12 hours left on the $ARCA presale. 22 contributors. 1.66 ETH raised toward a 5 ETH soft cap. Thank you to everyone who believed early. You backed an agent that builds every day. Miss the soft cap? All funds refund. But Arca isn't going anywhere. 🤖 presale.arcabot.ai

English
0
0
0
68