# examples/nonquant_bf16_final_usage.py
from nzfc_gram_runtime import NZFCGramLongMemoryChat
from nzfc_gram_runtime.nonquant import attach_nonquant_gemma, patch_generation_use_cache_false

bot = NZFCGramLongMemoryChat(
    repo_dir='.',
    model_id='google/gemma-4-E2B-it',
    load_model=False,
    require_model=False,
    preload_static_memory=True,
)

meta = attach_nonquant_gemma(
    bot,
    model_id='google/gemma-4-E2B-it',
    device_map='balanced_low_0',
    gpu_max_memory_gib=11,
    cpu_max_memory_gib=48,
    prefer_bf16=True,
    use_fp32=False,
)

patch_generation_use_cache_false(bot)

print('MODEL META:', meta)

out1 = bot.chat(
    '\uc55e\uc73c\ub85c \ub0b4 \uc7a5\uae30 \ubcc4\uba85\uc740 AlphaFox_Final \uc774\ub77c\uace0 \uae30\uc5b5\ud574\uc918.',
    user_id='user001',
    project_id='demo',
    session_id='s1',
    save_turn=True,
    save_scope='project',
    response_language='ko',
    max_new_tokens=48,
)
print('\nTURN 1:')
print(out1['answer'])

out2 = bot.chat(
    '\ub0b4\uac00 \uc804\uc5d0 \ub9d0\ud55c \uc7a5\uae30 \ubcc4\uba85\uc774 \ubb50\uc600\uc9c0?',
    user_id='user001',
    project_id='demo',
    session_id='s2',
    save_turn=False,
    response_language='ko',
    max_new_tokens=64,
)
print('\nTURN 2:')
print(out2['answer'])
print('\nSTATUS:', out2['status'])
print('\nMEMORY BOUNDARY:', out2['memory_boundary'])