dotcache/integrations/__init__.py

from .llama import (
    LlamaDotCacheHarness,
    LlamaDotCacheModelAdapter,
    LlamaReplayRecord,
    build_llama_page_trace_records,
    build_llama_prefill_page_trace_records,
    export_llama_page_traces,
    run_llama_generation_harness,
    run_llama_page_trace_capture_harness,
    run_llama_replay_harness,
    transformers_available,
)
from .qwen2 import (
    Qwen2DotCacheHarness,
    Qwen2DotCacheModelAdapter,
    run_qwen2_generation_harness,
    run_qwen2_loss_harness,
    run_qwen2_replay_harness,
)
from .qwen35 import (
    Qwen35AttentionSubsetDotCacheHarness,
    Qwen35AttentionSubsetDotCacheModelAdapter,
    Qwen35AttentionSubsetHarness,
    Qwen35AttentionSubsetModelAdapter,
    Qwen35DeltaNetStateRecord,
    Qwen35DeltaNetStateHarness,
    Qwen35DeltaNetStateModelAdapter,
    build_qwen35_deltanet_state_sample,
    capture_qwen35_deltanet_state_sample,
    Qwen35TextHarness,
    Qwen35TextModelAdapter,
    build_attention_subset_page_trace_records,
    build_attention_subset_prefill_page_trace_records,
    inspect_qwen35_deltanet_state,
    inspect_qwen35_hybrid_state,
    load_qwen35_text_only_from_pretrained,
    export_attention_subset_page_traces,
    run_qwen35_attention_subset_prefill_ablation_harness,
    run_qwen35_attention_subset_page_trace_capture_harness,
    run_qwen35_attention_subset_dotcache_harness,
    run_qwen35_attention_subset_replay_harness,
    run_qwen35_deltanet_state_ablation_harness,
    run_qwen35_text_generation_harness,
    run_qwen35_text_loss_harness,
    save_qwen35_deltanet_state_sample,
    summarize_qwen35_dotcache_fit,
    summarize_qwen35_hybrid_state,
)
from .vllm_adapter import (
    VllmAdapterConfig,
    VllmDotCacheModelAdapter,
    VllmPagedKVCache,
    VLLM_V1_MULTIPROCESSING_ENV,
    configure_vllm_inprocess_runtime,
    get_vllm_version,
    install_dotcache_on_vllm_model,
    install_dotcache_on_vllm_runtime,
    require_supported_vllm_version,
    vllm_available,
)

__all__ = [
    "LlamaDotCacheHarness",
    "LlamaDotCacheModelAdapter",
    "LlamaReplayRecord",
    "build_llama_page_trace_records",
    "build_llama_prefill_page_trace_records",
    "export_llama_page_traces",
    "Qwen2DotCacheHarness",
    "Qwen2DotCacheModelAdapter",
    "Qwen35AttentionSubsetDotCacheHarness",
    "Qwen35AttentionSubsetDotCacheModelAdapter",
    "Qwen35AttentionSubsetHarness",
    "Qwen35AttentionSubsetModelAdapter",
    "Qwen35DeltaNetStateRecord",
    "Qwen35DeltaNetStateHarness",
    "Qwen35DeltaNetStateModelAdapter",
    "build_qwen35_deltanet_state_sample",
    "capture_qwen35_deltanet_state_sample",
    "Qwen35TextHarness",
    "Qwen35TextModelAdapter",
    "build_attention_subset_page_trace_records",
    "build_attention_subset_prefill_page_trace_records",
    "VllmAdapterConfig",
    "VllmDotCacheModelAdapter",
    "VllmPagedKVCache",
    "VLLM_V1_MULTIPROCESSING_ENV",
    "configure_vllm_inprocess_runtime",
    "get_vllm_version",
    "install_dotcache_on_vllm_model",
    "install_dotcache_on_vllm_runtime",
    "require_supported_vllm_version",
    "inspect_qwen35_deltanet_state",
    "inspect_qwen35_hybrid_state",
    "load_qwen35_text_only_from_pretrained",
    "export_attention_subset_page_traces",
    "run_qwen35_attention_subset_prefill_ablation_harness",
    "run_qwen35_attention_subset_page_trace_capture_harness",
    "run_qwen35_attention_subset_dotcache_harness",
    "run_qwen35_attention_subset_replay_harness",
    "run_qwen35_deltanet_state_ablation_harness",
    "run_llama_generation_harness",
    "run_llama_page_trace_capture_harness",
    "run_llama_replay_harness",
    "run_qwen2_generation_harness",
    "run_qwen2_loss_harness",
    "run_qwen2_replay_harness",
    "run_qwen35_text_generation_harness",
    "run_qwen35_text_loss_harness",
    "save_qwen35_deltanet_state_sample",
    "summarize_qwen35_dotcache_fit",
    "summarize_qwen35_hybrid_state",
    "transformers_available",
    "vllm_available",
]