Delete start_E.sh

start_E.sh

@@ -1,45 +0,0 @@
-#!/bin/bash
-set -e
-MODEL_PATH="${MODEL_PATH:-/models/ggml-model-i2_s.gguf}"
-BITNET_PORT="${BITNET_PORT:-8080}"
-N_CTX="${N_CTX:-512}"
-N_THREADS="${N_THREADS:-2}"
-PORT="${PORT:-7860}"
-SERVER="server_E_v21"
-
-echo "[START] Cerebro E v21.5 (BitNet REAL)"
-[ ! -f "$MODEL_PATH" ] && echo "[START] FATAL: modelo no encontrado" && exit 1
-echo "[START] Modelo: $(du -h $MODEL_PATH | cut -f1)"
-
-BITNET_LOG="/tmp/bitnet.log"
-> "$BITNET_LOG"
-
-# Anti-alucinación v21.5:
-#   --cache-ram 0  → desactiva cache RAM (evita bug {{{{ de ik_llama.cpp)
-#   --jinja               → habilita chat template nativo Llama-3/ChatML
-#   --no-mmap             → evita lecturas corruptas por memory-map
-#   --ctx-size 512        → n_ctx pequeño para single-token output (<1ms evaluación)
-llama-server \
-    --model "$MODEL_PATH" \
-    --host 127.0.0.1 \
-    --port "$BITNET_PORT" \
-    --ctx-size "$N_CTX" \
-    --threads "$N_THREADS" \
-    --gpu-layers 0 \
-    --no-mmap \
-    --jinja \
-    --cache-ram 0 \
-    2>&1 | tee "$BITNET_LOG" | sed 's/^/[BITNET] /' &
-LLAMA_PID=$!
-
-MAX_WAIT=240; ELAPSED=0
-while [ $ELAPSED -lt $MAX_WAIT ]; do
-    kill -0 $LLAMA_PID 2>/dev/null || { echo "[START] FATAL: llama-server murió"; tail -20 "$BITNET_LOG"; exit 1; }
-    curl -sf "http://127.0.0.1:${BITNET_PORT}/health" > /dev/null 2>&1 && break
-    sleep 5; ELAPSED=$((ELAPSED+5))
-    [ $((ELAPSED%30)) -eq 0 ] && echo "[START] ${ELAPSED}/${MAX_WAIT}s cargando..."
-done
-[ $ELAPSED -ge $MAX_WAIT ] && echo "[START] FATAL: timeout" && exit 1
-echo "[START] BitNet ONLINE — iniciando FastAPI..."
-
-exec uvicorn ${SERVER}:app --host 0.0.0.0 --port "$PORT" --workers 1 --log-level info