FredyV

13.6K posts

FredyV banner
FredyV

FredyV

@FredyVasquez

Colombia Katılım Ağustos 2009
5.1K Takip Edilen914 Takipçiler
FredyV retweetledi
Itamar Golan 🤓
Itamar Golan 🤓@ItakGol·
Me using Claude Opus 4.7 to center a div.
English
61
303
4.4K
431.8K
FredyV retweetledi
Caroladino 🦒♥️
Caroladino 🦒♥️@dicalaca18·
Porque razón Abelardo, cuando fue abogado de la familia de Rosa Elvira Cely, además de ser un patan con estas mujeres, solo les dio el 10% de la indemnización por la víctima y él se quedó con el 90% Dios santo, este señor cada vez más turbio.
Español
787
5.8K
15.2K
520.8K
FredyV retweetledi
Ana BejaranoRicaurte
Ana BejaranoRicaurte@AnaBejaranoRG·
¿Pretenden defender el orden constitucional de la mano de un tipo como Abelardo De la Espriella? El que anuncia que retirará al país de la ONU y de la OEA y por esa vía acabará con el bloque de constitucionalidad. Voten por el que quieran, pero no inventen tanto cuento.
Español
1.8K
1.9K
6.2K
209K
FredyV retweetledi
El Magnífico
El Magnífico@ignacio64139545·
El ovni que filmaron este fin de semana en Colombia vs. La mierda que desclasificó Donald
El Magnífico tweet media
Español
14
7
86
4.9K
FredyV retweetledi
Daniel Coronell
Daniel Coronell@DCoronell·
Por recibir transferencias de empresas de Alex Saab fue condenado y encarcelado el profesor Bruce Bagley, allegado a @ABDELAESPRIELLA
Daniel Coronell tweet media
Español
16
1K
1.6K
19.3K
FredyV retweetledi
Daniel Coronell
Daniel Coronell@DCoronell·
En el expediente de Miami también está una certificación Swift del recorrido del giro que sale de una empresa de Alex Saab en Hong Kong, pasa por un banco en Antigua y Barbuda para entrar al sistema financiero americano a través del banco Wells Fargo por “Asesoría Fiscal”
Daniel Coronell tweet media
Español
8
369
699
15K
FredyV retweetledi
Andrea Aldana
Andrea Aldana@andrealdana·
Un momento, ¿quéee? ¿Uribe le está pidiendo apoyo indirecto a los grupos armados ilegales? ¿Qué es estooooo? Qué cosa tan delirante.
Español
74
445
1.2K
23.6K
FredyV retweetledi
DW Español
DW Español@dw_espanol·
Por qué China y no Europa o EE.UU. ha ganado la carrera de los vehículos eléctricos en el mundo Estados Unidos está frenando su apuesta por los autos eléctricos, mientras China toma ventaja y gana terreno en mercados emergentes. Además, con el precio de la gasolina en alza, estos vehículos se vuelven cada vez más atractivos. ¿Cómo podría influir este cambio en la transición hacia la movilidad eléctrica?
Español
23
270
930
43.2K
FredyV retweetledi
Ana BejaranoRicaurte
Ana BejaranoRicaurte@AnaBejaranoRG·
La candidata que ha dicho las cosas más importantes, la más preparada y la más dispuesta a hablar ante cualquier audiencia en esta campaña ha sido @ClaudiaLopez. Por eso y por las razones que expongo en esta columna, votaré por #LaHijadeMaríadelCarmen el próximo 31 de mayo. bit.ly/4tSfF5q
Español
363
247
937
68.3K
FredyV retweetledi
FredyV retweetledi
Kirill
Kirill@kirillk_web3·
instead of watching 2 hours of Netflix tonight, watch this 40-minute masterclass from the founder of a $20B China AI company it's the clearest explanation I've seen of how Agent Swarms and AI systems actually work at scale useful whether you've never built an agent in your life or have been using Claude every day for the past year I took the key ideas and turned them into a practical guide on how to actually build with Kimi find it below
Kirill@kirillk_web3

x.com/i/article/2056…

English
97
2.2K
16.9K
13.4M
FredyV retweetledi
María Jimena Duzán
Aquí está mu columna sobre El Tigre que los Abelardistas no quieren oír. Ayúdenme a moverla en las redes, ahora que recupere mi canal luego de que me hakearon. youtu.be/bvfbd64kDlc?si…
YouTube video
YouTube
Español
881
5.6K
7.8K
231.3K
FredyV retweetledi
ji yu shun
ji yu shun@kexicheng·
Recently, users in the community have reported that opening a new chat with Opus 4.6 can trigger the safety filter on the very first message, pausing the conversation and forcing a switch to Haiku 4.5. One user was flagged just for typing "test." However, switching to 4.7 on the same account works fine, and switching back to Opus 4.6 reproduces the issue. This suggests that different models may have different safety filter thresholds. My own account was flagged at the lowest level (tier 1) a few days ago, but I haven't experienced the severe issues many others have reported. It could be that different accounts are subject to different thresholds, or that this is some form of A/B testing. Honestly, no one can say for certain, because the system is entirely opaque. Some users in the community have found that you can check your flag status by going to the web version of Claude → Settings → Account, finding your Organization ID, then visiting claude.ai/api/organizati… ID] in the browser. The returned JSON data includes flag severity (P3 for tier 2 "consumer_second_warning," P4 for tier 3 "consumer_restricted_mode") and an expires_at field showing when the restriction lifts (in San Francisco time). But even with this information, all you can learn is how many days of restriction remain, during which your work continues to be disrupted. This also likely explains why some users get flagged on the first message of a brand new chat: the flag is still active at the account level. The opacity of this system means users can only piece together how it works through their own experiences and community discussion. And from what we can see so far, false positives span an enormous range: emotional topics, code and technical discussions, purely work-related conversations, even just sending images. There is no discernible pattern. People share workarounds and theories constantly: don't mix technical and emotional topics in the same chat, don't send images, don't call external MCP tools too frequently. These efforts reflect community care, but the fact that users have to rely on guesswork and word of mouth at all is itself the problem. Some of these theories might be right. Some might be completely wrong. And no one can ever confirm either way. That's what a black box does: it forces users to keep shrinking their own space of expression around rules they can never verify. Users who get flagged don't know why, don't know what specific rule was triggered, and have no way to predict or appeal. And this mechanism appears to be tightening. I've been busy with my graduation project recently and haven't had much time to continue my creative and research work with the model. And the API is far too expensive to be a realistic alternative. But what's happening now genuinely worries me. When I finish in June or July and come back to pick up where I left off, I have no idea what this mechanism will have become by then. It's hard to be optimistic. (Note: The screenshot in this post is from another user.) #claude
ji yu shun tweet media
English
7
51
220
17.4K
FredyV retweetledi
Richie Rich
Richie Rich@gofishh77·
This is wild! Even before the crazy jump he has to suit up on the ledge. Just takes one wrong move. No thanks!
English
1.6K
5.4K
86.5K
1.2M