home.social

Search

1000 results for “sen”

  1. Tensions between North Korea and South Korea are running high this week after Pyongyang sent a letter to the U.N. Security Council repeating accusations that Seoul fabricated a September land mine incident. japantimes.co.jp/news/2026/10/ #asiapacific #northkorea #southkorea #asia #eastasia

  2. europesays.com/africa/490260/ Algeria to send troops to Niger in first foreign deployment in over 50 years #Algeria

  3. #job in #Bremen
    Referent:in (w/m/d) Oberflächengewässer / chemischer Gewässerzustand - Freie Hansestadt Bremen - Senatorin für Umwelt, Klima und Wissenschaft
    #FediHire

    https://www.greenjobs.de/a100155539

  4. Jon Russell from Asia Tech Review says the valuations don’t make sense. AI startups are being priced beyond Apple or Amazon, yet we haven’t seen any filings. When OpenAI finally goes public we’ll finally see what’s behind the curtain.

    Watch on YouTube 👇
    youtube.com/shorts/eBnbM50IKmY

  5. 🧿 El cantante Justin Bieber difunde un video celebrando la victoria de #Lula con la señal de la L y fumando un habano cubano.

    Varios artistas internacionales se sumaron en apoyo a Lula y contra el avance del fascismo en Latinoamérica.

    #BoraLula #Brasil #JustinBieber

  6. europesays.com/africa/489893/ Algeria to send troops to Niger in first foreign deployment in over 50 years #Algeria

  7. You review a 30-line template diff. Production sends 40,000 tokens assembled at request time from six sources, four of which changed this week and none of which appear in the pull request. The diff you approved is not the change you shipped.

    This is the part of LLMOps that keeps biting teams after they have evals, tracing, and a gateway in place. The template is versioned in git. The context is assembled by code paths nobody treats as release artifacts.

    The prompt is a build output


    List what actually goes into a request for a typical RAG-plus-tools system:

    • The system template (in git, reviewed).
    • Top-k retrieved chunks from an index that gets rebuilt on its own schedule.
    • Tool JSON schemas, whose descriptions are prompt text written by whoever last touched the tool.
    • A conversation summary produced by another model call.
    • A user profile or tenant config blob.
    • Few-shot examples chosen at runtime by a selector.

    Each is a dependency with its own version and its own deploy cadence. The retrieved chunks change when someone reindexes. The tool schema changes when a backend engineer renames a parameter and rewrites the description. Neither triggers a review of the prompt.

    So when quality drops on Tuesday and no code shipped, there is no diff to look at. That is the failure this whole post is about.

    Fingerprint everything the model sees


    Compute a content hash over the inputs that can change model behaviour, and log it with every trace:

    • model id and version
    • template version
    • retrieval snapshot id (the index build, not the query)
    • the fully rendered prompt
    • decoding parameters that matter

    Store fingerprints always. Store full rendered prompts sampled or redacted, depending on your data rules — full-text retention is a compliance decision, not an observability one. The fingerprint is small, cheap, and answers "what changed" during an incident in one query.

    The trade-off: a fingerprint that includes a timestamp, a request id, or a per-user greeting has near-zero collision rate and zero diagnostic value. Hash the parts that affect generation, not the metadata around it. I have watched a team spend a day on "why is our prompt hash changing every request" when the answer was a locale string.

    A small assembler with provenance and eviction


    You do not need a framework for this. You need three properties: every section carries a trust level, every section carries an eviction priority, and over-budget on a pinned section is a hard error rather than a silent trim.

    from dataclasses import dataclass, field
    from hashlib import sha256
    import re

    @dataclass
    class Section:
    name: str
    text: str
    trusted: bool # False for retrieval, tool output, user uploads
    priority: int # higher survives truncation
    pinned: bool = False # never evict; over budget means fail loudly

    INSTRUCTION_LIKE = re.compile(
    r"(ignore (all )?(previous|prior) instructions|you are now|system prompt)",
    re.IGNORECASE,
    )

    def screen(text: str) -> str:
    # Cheap heuristic, not a guarantee. Log hits and route to a stricter path.
    return INSTRUCTION_LIKE.sub("[filtered]", text)

    def count_tokens(s: str) -> int:
    # Swap in tiktoken/the provider counter. One counter per model family.
    return len(s) // 4

    @dataclass
    class Assembled:
    prompt: str
    fingerprint: str
    dropped: list[str] = field(default_factory=list)

    def assemble(sections, budget, model_id, template_version, snapshot_id):
    ordered = sorted(sections, key=lambda s: (s.pinned, s.priority), reverse=True)
    kept, dropped, used = [], [], 0
    for s in ordered:
    body = f"<{s.name} trust={'high' if s.trusted else 'low'}>\n"
    body += (s.text if s.trusted else screen(s.text)) + f"\n</{s.name}>"
    cost = count_tokens(body)
    if used + cost > budget:
    if s.pinned:
    raise ValueError(f"pinned section {s.name} exceeds budget")
    dropped.append(s.name)
    continue
    kept.append((s, body))
    used += cost
    # Render in original order so layout is stable across requests.
    pos = {id(s): i for i, s in enumerate(sections)}
    kept.sort(key=lambda pair: pos[id(pair[0])])
    prompt = "\n\n".join(body for _, body in kept)
    h = sha256()
    for part in (model_id, template_version, snapshot_id, prompt):
    h.update(part.encode())
    return Assembled(prompt, h.hexdigest()[:16], dropped)

    # Usage: build once per request, log the fingerprint beside the trace id.
    ctx = assemble(
    sections=[
    Section("question", "why was my claim denied?", trusted=True, priority=10, pinned=True),
    Section("policy_chunk", retrieved_text, trusted=False, priority=5),
    Section("profile", tenant_blob, trusted=True, priority=1),
    ],
    budget=12_000,
    model_id="claude-sonnet-4",
    template_version="support-v14",
    snapshot_id=index_build_id,
    )

    Two details in there are load-bearing. Sections render in original insertion order rather than priority order, because reordering the same content changes model behaviour and makes your eval deltas unattributable. And pinned turns budget pressure into an exception instead of a degraded#ai #llm #devops #python #software #coding #development #engineering #inclusive #community
    Context Assembly Is a Build Step: Versioning the Prompt You Actually Send

  8. Trump says El-Sayed should have to pass through metal detector to enter Senate
    #Trump

    Trump says El-Sayed should hav...

  9. Le Pen casts herself as protector of France as troubles swirl – POLITICO

    According to a senior National Rally official, Le Pen wanted to use her speech to cast herself in…
    #Europe #EU #Borders #Budget #Crisis #debt #EdouardPhilippe #Elections #EmmanuelMacron #EuropeanUnion #Eurozone #Farleft #Farright #France #Jean-LucMélenchon #JordanBardella #MarineLePen #Markets #poll #Rights #Security
    europesays.com/europe/166443/

  10. We handed the agent a webhook client that signs each outbound payload with HMAC before sending. It read the integration guide, built the serializer, and the local tests passed against a stub that accepted any well-formed body. Before opening the PR, it decided the body looked messy: the keys arrived in the order the fields were populated, so it ran a recursive sort to produce alphabetical, deterministic JSON.

    The tests still passed. CI went green. The first production delivery returned 401 signature_invalid, intermittently — only for some payloads, because the receiver computed its digest over the exact bytes it received and expected the same field order the sender used in the signing step. The agent had treated object key order as cosmetic; for a signed payload, it is part of the contract.

    Why the tests stayed green


    The test asserted that the body parsed and that the HMAC function was called with some secret. It never compared the digest against the exact serialized bytes the receiver would rebuild, and it never exercised the real signing serializer against a verifying server. A mock that only checks shape cannot tell you whether the bytes match, so a reformatted body sailed through.

    What actually broke


    JSON objects are unordered semantically: a parser treats {a, b} and {b, a} as the same value. HMAC does not compare values, it hashes bytes. When the sender signs the body before sorting, then sorts before sending, the receiver rehashes the reordered bytes and gets a different digest. Same payload, same secret, valid structure — just a different byte order — fails verification. The sort happened between signing and transmission, which is exactly the window that has to stay byte-stable.

    Keep the signing step and the wire bytes identical


    • Sign the exact serialized bytes you will send. Do not sign an in-memory object and transmit a reformatted copy.
    • If deterministic ordering matters for verification, define one canonical serialization (sorted keys, fixed separators, no added whitespace) and apply it before the digest, on both sender and receiver. Never reformat after signing.
    • Keep pretty-printing for logs only. Write the normalized debug copy to the log, and transmit the signed raw bytes.


    Tests that catch this


    1. Sign a fixed payload, transmit it unchanged, and verify the digest recomputed from the received bytes equals the stored one.
    2. Reject any transform — sorting, pretty-printing, whitespace normalization — that runs after the signing call in the call stack.
    3. Replay one signed webhook against an environment that enforces the signature, not only a stub that parses the JSON.


    Where the contract lives


    When the signed shape and field order are stated in the integration spec, that spec should stay the source of truth locally rather than a hand-kept serialization note. This is the local-first workflow we built around in Powerduck: keep the contract on disk, validate the actual outgoing bytes against it, and replay one real request against an enforcing endpoint instead of trusting a permissive mock.

    Before you call the signing path done:

    1. Was the digest computed over the exact bytes transmitted?
    2. Did any sort, pretty-print, or normalize run after signing?
    3. Did a real verifying endpoint accept the signature, or only a shape-accepting stub?
    4. Is the expected serialization order written in the contract, not assumed?

    A normalized, pretty body reads better in review. On the wire between two parties that sign, byte stability is the product. Order is free inside a parser; it is expensive after a digest.#aiagents #aicode #apicontract #security #software #coding #development #engineering #inclusive #community
    The agent sorted the JSON body keys to tidy it up, and the request signature stopped matching

  11. Audit your contractors' patch management procedures and apply Oracle PeopleSoft updates before ShinyHunters sends you a thank-you card.

    Reward: You've received a Sponsored Maintenance Notice — proudly displayed on your recruitment portal for days!

    en.wikipedia.org/wiki/2026_FBI

    #Cybersecurity #DataBreach #FBI #ShinyHunters #Oracle #Ransomware (3/3)

  12. 📢 We're Hiring! | Senior Python Developer (Python)

    Remote contract, pay discussed during screening
    ✅ Writing Python that solves prompts in SFT datasets
    ✅ Ranking paired model outputs for RLHF feedback
    ✅ Testing depth across unit, integration and property-based suites
    How do you decide when a refactor is safe to ship?

    👉 Apply here: nearskill.in/jobs/senior-pytho

    #JobSearch #Python #RemoteJobs #AIJobs

  13. Ignacio Mier defiende ley de pagos digitales contra lavado de dinero

    El senador aseguró que la legislación busca combatir operaciones ilícitas y aclaró que no prohíbe los pagos en efectivo en pequeños comercios; la reforma promueve la digitalización financiera y la formalización económica.

    Por Gabriela Díaz | Reportera                                           

    El senador Ignacio Mier Velazco defendió la Ley de Economía Digital para Pagos Digitales y Electrónicos, al afirmar que busca combatir el lavado de dinero, cerrar espacios al crimen organizado y fortalecer la formalización económica, sin eliminar el uso de efectivo en las transacciones cotidianas.

    El legislador sostuvo que la legislación no pretende perjudicar a la población ni afectar a los pequeños establecimientos comerciales, sino establecer mecanismos para enfrentar las operaciones ilícitas y el aprovechamiento de vacíos regulatorios con fines de evasión fiscal y lavado de recursos.

    Pagos en efectivo

    Mier Velazco rechazó las versiones de que la nueva legislación impedirá realizar pagos en efectivo y puso como ejemplo las compras en carnicerías, tiendas de abarrotes, puestos de dulces y mercados, donde, aseguró, las personas podrán continuar utilizando billetes y monedas.

    El senador argumentó que las disposiciones están dirigidas a combatir actividades delictivas y no a restringir las formas de pago de las familias mexicanas. También cuestionó a quienes, desde su perspectiva, presentan la reforma como una medida que afectará a los sectores de menores ingresos.

    “Es falso”, expresó al referirse a las versiones de que únicamente se permitirá pagar con tarjeta, y defendió la necesidad de distinguir entre las operaciones comerciales ordinarias y aquellas que puedan estar relacionadas con actividades ilícitas.

    Digitalización financiera

    La legislación, aprobada por el Senado de la República el martes 20 de octubre, de acuerdo con la información proporcionada, plantea impulsar la digitalización de los medios de pago para reducir las barreras de acceso a la economía formal y ampliar la inclusión financiera.

    Entre sus objetivos se encuentra facilitar el acceso a servicios financieros para sectores históricamente excluidos, entre ellos mujeres, habitantes de localidades rurales y personas con rezago educativo.

    La incorporación de más personas y negocios al sistema financiero formal puede facilitar el acceso a instrumentos de ahorro, crédito y seguros, además de generar condiciones para que las pequeñas unidades económicas acrediten sus actividades y tengan mayores posibilidades de obtener financiamiento.

    Formalización económica

    El desarrollo de sistemas electrónicos de pago también busca reducir los tiempos y costos de las transacciones entre personas y empresas, así como facilitar la participación de micro y pequeñas empresas en la economía formal.

    La digitalización puede contribuir a mejorar la trazabilidad de determinadas operaciones financieras, aunque sus resultados dependen de la implementación de la legislación, la supervisión de las autoridades y las garantías para proteger los datos de los usuarios.

    La discusión sobre la reforma incluye, por tanto, el combate a las operaciones ilícitas, la ampliación de los servicios financieros y la preservación de alternativas de pago accesibles para la población. –sn–

    Sociedad Noticias

    ¡Conéctate con Sociedad Noticias! Suscríbete a nuestro canal de YouTube y activa las notificaciones, o bien, síguenos en las redes sociales: Facebook, Twitter e Instagram.

    También, te invitamos a que te sumes a nuestro canal de información en tiempo real a través de Telegram.

    #NoticiasMX #PeriodismoParaTi #PeriodismoParaTiSociedadNoticias #Cdmx #economíaDigital #economíaFormal #IgnacioMier #inclusiónFinanciera #Información #InformaciónMéxico #LavadoDeDinero #México #Morena #noticia #noticias #NoticiasMéxico #NoticiasSociedad #pagosDigitales #pagosElectrónicos #SenadoDeLaRepública #SN #Sociedad #SociedadNoticias #SociedadNoticiasCom #sociedadNoticias #SociedadNoticiasCom
  14. AG Todd Blanche says DOJ is "doing our own research under the law" with Alaska Senate challenge (Kaia Hubbard/CBS News)

    cbsnews.com/news/todd-blance-a
    memeorandum.com/261025/p50#a26

  15. NOW PLAYING == On Sunday with Jack Cocchiarella 10/18/26: Jack and Sen. Bernie Sanders (I-Vt.), Rep. Angie Nixon (D-Fla.), Rep. Ro Khanna (D-Calif.), and Sen. Chris Van Hollen (D-Md.) discuss everything from the Trump's midterm blitz to the 2028 Democratic primary.

    youtube.com/watch?v=i6Aa8ETvoD0

    #news #uspol #politics #trump #commentary #progressive #left #JackCocchiarella #MeidasTouch #Democrats #vsn #radio #DiverseSpectrumOfTheLeft #SupportIndependentMedia #Resist

  16. Lula and Flavio Bolsonaro cast their ballot in presidential runoff

    President Lula and right-wing Senator Flavio Bolsonaro cast ballots in Brazil's presidential runoff in a vote that could shift the politics of Latin America's largest nation. #News #Reuters #Newsfeed #Brazil #Lula #FlavioBolsonaro #JairBolsonaro #BrazilElection #RightWing #BreakingNews #WorldNews #Election2026 #BrazilPolitics #PresidentialRunoff Read the story here: 👉 Subscribe: Keep up with the latest…

    fllics.com/zh/video/lula-and-f

  17. Personalbericht – Personalbestand der Stadt Hamburg leicht gestiegen – Politik

    Hamburg (dpa/lno) – Trotz Bemühungen des Senats, die Zahl der städtischen Beschäftigten konstant zur Bevölkerung Hamburgs zu halten,…
    #Hamburg #Deutschland #Deutsch #DE #Schlagzeilen #Headlines #Nachrichten #News #Europe #Europa #EU #Bildung #hamburg #InnereSicherheit #Politik #Rathaus #Raum151 #Senat #Sicherheitspolitik #SüddeutscheZeitung #Verwaltung
    europesays.com/de/1365396/

  18. Blanche says Trump isn’t trying to meddle with Alaska election results by suing over Senate race

    WASHINGTON – Attorney General Todd Blanche on Sunday defended the Trump administration’s lawsuit to block certification of Alaska’s…
    #UnitedStates #US #USA #DanSullivan #DanaBash #DonaldTrump #elections #LisaMurkowski #MaryPeltola #Politics #ToddBlanche #trump #Washingtonnews
    europesays.com/3331387/

Share on Mastodon

Enter the server where you have an account.