LICENSE
MANIFEST.in
README.md
install.sh
manage.sh
pyproject.toml
release.sh
run.sh
uninstall.sh
benchmarks/__init__.py
benchmarks/offline_ab.py
configs/agentx-qwen35-frontier-mtp2.example.json
configs/agentx-qwen35-frontier-target-only.example.json
configs/qwen25-14b-05b-adaptive-b128.toml
configs/qwen25-14b-05b-arc-easy.toml
configs/qwen25-14b-05b.toml
configs/qwen25-14b-eagle-arc-easy.toml
configs/qwen25-14b-eagle-relaxed.toml
configs/qwen25-14b-eagle.toml
configs/qwen3-8b-eagle3.toml
configs/qwen35-35b-a3b-eagle3-adaptive-fixed5-isolation.toml
configs/qwen35-35b-a3b-eagle3-adaptive.toml
configs/qwen35-35b-a3b-eagle3.toml
configs/qwen35-35b-a3b-frontier-mtp2-adaptive.toml
configs/qwen35-35b-a3b-frontier-mtp2-b16.toml
configs/qwen35-35b-a3b-frontier-mtp2.toml
configs/qwen35-35b-a3b-mtp2.toml
docs/agentx_256k_benchmark.md
docs/arc_easy_regression.md
docs/default_gamma1to4_regression_20260907.md
docs/draft_b16_optimization_checkpoint_20260915.md
docs/draft_b16_recovery_handoff_20260914.md
docs/draft_latest_stack_performance_recovery_20260904.md
docs/dynamic_gamma_gsm8k_b128_optimization_report.md
docs/eagle_b16_graph_benchmark_commands.md
docs/eagle_dynamic_gamma_0to4_gsm8k_b128_report.md
docs/gsm8k_b128_recovery_and_arc_easy_20260913.md
docs/latest_stack_gsm8k_graph_performance_20260904.md
docs/latest_stack_validation_20260903.md
docs/migration_status.md
docs/qwen35_35b_a3b_mtp2_eagle3_graph_eager_c16_report.md
docs/qwen35_35b_a3b_mtp2_gsm8k_report.md
docs/qwen35_eagle3_dynamic_gamma_c1_c16_report.md
docs/qwen35_frontier_mtp2.md
docs/qwen35_frontier_mtp2_gsm8k_commands.md
docs/qwen35_speedbench_coding_al.md
docs/vspec_adaptive.md
docs/vspec_adaptive_npu_report_20260902.md
docs/vspec_online_adaptive_npu_report_20260902.md
profiles/qwen25_draft_gamma0_sync_transition.json
profiles/qwen25_draft_gsm8k_graph_b128_g4.json
profiles/qwen25_draft_gsm8k_graph_g3_control.json
profiles/qwen25_draft_target_only_control.json
profiles/qwen25_eagle_gsm8k_graph_b128.json
profiles/qwen25_eagle_gsm8k_graph_g2_control.json
scripts/benchmark_qwen35_c1_c16.sh
scripts/benchmark_qwen35_frontier_mtp2_gsm8k.sh
scripts/benchmark_qwen35_mtp2_gsm8k.sh
scripts/measure_qwen35_speedbench_al.sh
scripts/prepare_speedbench_coding.py
scripts/validate_qwen35_long_context_retrieval.py
scripts/verify_release.py
src/vllm_hust_vspec/__init__.py
src/vllm_hust_vspec/__main__.py
src/vllm_hust_vspec/_version.py
src/vllm_hust_vspec/cli.py
src/vllm_hust_vspec/compatibility.py
src/vllm_hust_vspec/config.py
src/vllm_hust_vspec/host_compat.py
src/vllm_hust_vspec/model_store.py
src/vllm_hust_vspec/online_benchmark.py
src/vllm_hust_vspec/patches.py
src/vllm_hust_vspec.egg-info/PKG-INFO
src/vllm_hust_vspec.egg-info/SOURCES.txt
src/vllm_hust_vspec.egg-info/dependency_links.txt
src/vllm_hust_vspec.egg-info/entry_points.txt
src/vllm_hust_vspec.egg-info/requires.txt
src/vllm_hust_vspec.egg-info/top_level.txt
src/vllm_hust_vspec/adaptive/__init__.py
src/vllm_hust_vspec/adaptive/controller.py
src/vllm_hust_vspec/adaptive/entropy.py
src/vllm_hust_vspec/adaptive/fitting.py
src/vllm_hust_vspec/adaptive/online.py
src/vllm_hust_vspec/adaptive/profile.py
src/vllm_hust_vspec/adaptive/profiling.py
src/vllm_hust_vspec/adaptive/refill.py
src/vllm_hust_vspec/adaptive/runtime.py
src/vllm_hust_vspec/backends/__init__.py
src/vllm_hust_vspec/backends/draft.py
src/vllm_hust_vspec/backends/draft_body_quant.py
src/vllm_hust_vspec/backends/draft_hybrid.py
src/vllm_hust_vspec/backends/draft_parallel_update.py
src/vllm_hust_vspec/backends/draft_repetition.py
src/vllm_hust_vspec/backends/draft_vocab.py
src/vllm_hust_vspec/backends/eagle.py
src/vllm_hust_vspec/backends/eagle3_gdn_metadata.py
src/vllm_hust_vspec/backends/eagle3_hybrid.py
src/vllm_hust_vspec/backends/eagle3_replicated_sample.py
src/vllm_hust_vspec/backends/eagle_body_quant.py
src/vllm_hust_vspec/backends/eagle_draft.py
src/vllm_hust_vspec/backends/eagle_graph.py
src/vllm_hust_vspec/backends/eagle_host.py
src/vllm_hust_vspec/backends/eagle_metadata.py
src/vllm_hust_vspec/backends/eagle_parallel_update.py
src/vllm_hust_vspec/backends/eagle_rejection.py
src/vllm_hust_vspec/backends/eagle_runtime.py
src/vllm_hust_vspec/backends/eagle_target.py
src/vllm_hust_vspec/backends/eagle_tensor_fia.py
src/vllm_hust_vspec/backends/eagle_tp.py
src/vllm_hust_vspec/backends/kv_cache.py
src/vllm_hust_vspec/backends/mamba_compat.py
src/vllm_hust_vspec/backends/mtp.py
src/vllm_hust_vspec/backends/target_body_quant.py
src/vllm_hust_vspec/manifests/__init__.py
src/vllm_hust_vspec/manifests/vllm-hust-extension-v0.2.json
src/vllm_hust_vspec/models/__init__.py
src/vllm_hust_vspec/models/qwen2_eagle.py
tests/test_adaptive.py
tests/test_agentx_assets.py
tests/test_benchmark.py
tests/test_eagle3_gdn_metadata.py
tests/test_eagle3_hybrid.py
tests/test_eagle3_replicated_sample.py
tests/test_eagle_metadata.py
tests/test_eagle_parallel_update.py
tests/test_eagle_tp.py
tests/test_extension_lifecycle.py
tests/test_host_compat.py
tests/test_launcher.py
tests/test_mamba_compat.py
tests/test_management.py
tests/test_manifest.py
tests/test_model_store.py
tests/test_models.py
tests/test_mtp_incremental.py
tests/test_online_benchmark.py
tests/test_refill.py
tests/test_release.py