Akiragpu / modules /config.py

Commit History

  1. local-gpu: Qwen2.5-7B-Instruct 4-bit (~5GB, faster load, smarter)

    Isaac Quarenta
    dc575e5
  2. local-llm: swap 8B 16GB for Qwen2.5-3B (~6GB) — loads in ~10s on ZeroGPU FREE

    Isaac Quarenta
    89bd0bd
  3. persona: tom sério/curto estilo Kiami

    Isaac Quarenta
    b6444da
  4. fix(app): eliminar crash ChatbotDataMessages 'ping' + spam asyncio; NVIDIA default deepseek-v4.1-flash⁠

    Isaac Quarenta
    cf29563
  5. feat(providers): NVIDIA NIM como provider extra (build.nvidia.com)⁠

    Isaac Quarenta
    c487576
  6. feat(botcore): PV simples espelha o index (buttonId, midia, filtros) + sync api/config⁠

    Isaac Quarenta
    42306bf

  1. fix(persona): tom serio-coloquial-logico-racional + fim de respostas com aspas⁠

    Isaac Quarenta
    52cd142
  2. fix(persona): tom serio-coloquial-racional + declaracoes de afeto + strip de aspas em loop

    Isaac Quarenta
    e705438
  3. fix(chat+images): schnell primeiro no HF + interjeicoes/identidade sem web search

    Isaac Quarenta
    02995ae

  1. perf(boot)+fix(gpu): NLP/embeddings forcados a CPU, emotion trainer lazy, Docker removido do Akiragpu⁠

    Isaac Quarenta
    3c0fde3
  2. fix(gpu): forca 100% GPU - _nlp_device=0, embeddings cuda, +diffusers/torchvision p/ FLUX+PhiVision

    Isaac Quarenta
    8e2acc2
  3. feat(vision): Phi-3.5 Vision + FLUX local GPU (prefetch no build, lazy no boot) - fixa travamento

    Isaac Quarenta
    0951d77
  4. feat(gpu): GPU ativo por omissao (40s) + modelo local 8B com prefetch para /data⁠

    Isaac Quarenta
    69df760

  1. fix(chat): chat nunca parte no Gradio (fallback GPU->sem GPU em 429 ZeroGPU) + NLP/classificadores forcados a CPU + probes de CUDA reais

    Isaac Quarenta
    e48cf8f
  2. Switch to uncensored GPU local LLM and enable heavy model preference

    Isaac Quarenta
    bc7babe
  3. Enable explicit CUDA device handling for pipelines

    Isaac Quarenta
    9eb0b22
  4. feat(gpu): migracao para Space Gradio GPU (Akiragpu/L4) - modelo local 4-bit em 1 prioridade da chain, QLoRA real, app.py Chat/Treino/Status

    Isaac Quarenta
    bfcd5c0

  1. fix(contexto): dedup de contexto, isolamento por conversa e placeholder de busca⁠

    Isaac Quarenta
    04bb7b4

  1. feat(jev): System-Two deliberativo real — injecao Kahneman + consumidor jev_analysis + proactive/hostility no preclassify + debug system_two

    Isaac Quarenta
    3be1ac4

  1. fix(jev): aponta Defapi typesafe/jev-1.13 — /decisions + api.defapi.org (era /systemone em typesafe.ai)

    Isaac Quarenta
    456a4e7
  2. feat(jev): multi-key rotation (JEV_API_KEY/_2) + /debug/jev-status live ping

    Isaac Quarenta
    8bb3c4f
  3. feat(jev): add low-level noul/score/choice methods + jev_questions integration

    Isaac Quarenta
    3276d31
  4. feat(jev): System One Model integration — ultra-fast intent/emotion/routing/moderation

    Isaac Quarenta
    a2faf22

  1. Update modules/config.py

    Verified signature
    df3d3b0
  2. Update modules/config.py

    Verified signature
    4fd8b04

  1. feat(tokens/pdf): max token largo geral + pdf extra + formatacao rica prompt-only⁠

    Isaac Quarenta
    1e9d829

  1. fix(security): tag_everyone liberado p/ dono via LID/sender_jid⁠

    Isaac Quarenta
    4f3d10c

  1. fix: resposta curta reforçada + CoT sabe reply_author (reply_to_bot)

    Isaac Quarenta
    9bff8a5
  2. fix: modelos igual à Kiami — OpenRouter gpt-oss-20b, HF v0.3, CoT gemini-3.5-flash-lite

    Isaac Quarenta
    c8e9e6f
  3. feat(prompt): Akira ativa — conduz conversa, pergunta, discorda, defende/revoga

    Isaac Quarenta
    4b821e7
  4. fix: TokenRa provider + modelos corrigidos⁠

    Isaac Quarenta
    167c0ac

  1. fix: GROQ_MODEL → llama-3.3-70b-versatile (131K ctx, 280 t/s)⁠

    Isaac Quarenta
    fa33d08
  2. fix: OPENROUTER_MODEL → nvidia/llama-3.1-nemotron-70b-instruct:free (70B, 131K, free)

    Isaac Quarenta
    4e721ed
  3. fix: openrouter nemotron→gemini-2.0, mistral timeout 20s+connect, groq payload truncate, factual question gate⁠

    Isaac Quarenta
    628112f
  4. fix: remove toda lógica de truncar — prompt é a única fonte de verdade⁠

    Isaac Quarenta
    bc11537
  5. fix: persona rework + CoT skip + Groq payload + HF logging⁠

    Isaac Quarenta
    a1cb65a
  6. fix: provider chain + web search + gemini models⁠

    Isaac Quarenta
    46e7b44
  7. fix: BPE spacing (26 word fixes) + tighter personality (colder, shorter, Kiami-style)

    Isaac Quarenta
    e79f50a

  1. fix: Gemini 2.5-flash, Groq 7500 char skip, COT ENFORCE natural expansion, web_search opinion block

    Isaac Quarenta
    b6a0c21
  2. fix: respostas longas + delírio + fulano vs sicrano (terceira pessoa)⁠

    Isaac Quarenta
    68d1f2f

  1. fix: timeout 180s + opinião sem busca + Gemini 404⁠

    Isaac Quarenta
    0bc2344
  2. fix: natural fluxo dormiu vs tou bem + compact naturalidade⁠

    Isaac Quarenta
    4a8f28e

  1. fix: tradução COT ENFORCE bypass completo + coaching overlap⁠

    Isaac Quarenta
    0981e6a

  1. perf: CoT prompt SYNTHESIS condensado de 50 para 10 linhas, max_tokens volta a 2000

    Isaac Quarenta
    83f4f94
  2. perf: otimizações de performance - 2 workers, CoT skip trivial, MAX_ROUNDS=1, tools por relevância, blacklist 1h

    Isaac Quarenta
    8d04614

  1. fix: lugar=busca contextual; orroh? em vez de 'Diz lá'; sem listas complicadas

    Isaac Quarenta
    31161d2
  2. fix: CoT defensivo resolvido; contexto follow-up; pronomes resolvidos

    Isaac Quarenta
    33337c7
  3. fix: pesquisa autónoma com palavras reservadas; ignora definição de palavras; força uso dos resultados

    Isaac Quarenta
    cb8b220
  4. fix: pesquisa autónoma sem confirmação; apresenta como conhecimento próprio

    Isaac Quarenta
    46efd0f

  1. fix: CoT preservado, contradições resolvidas, prompt coerente⁠

    Isaac Quarenta
    c5f078a