mac-mlx-1c-1779513712.junit.xml

<?xml version="1.0" encoding="utf-8"?><testsuites name="pytest tests"><testsuite name="pytest" errors="0" failures="0" skipped="0" tests="196" time="195.879" timestamp="2026-05-23T13:21:55.007094+08:00" hostname="fluffy314s-Mac-mini.local"><testcase classname="tests.core.test_baseline" name="test_baseline_loads" time="1.995" /><testcase classname="tests.core.test_baseline" name="test_baseline_rejects_nonpositive_max_new_tokens" time="0.001" /><testcase classname="tests.core.test_baseline" name="test_baseline_generates_and_grows_kv" time="1.054" /><testcase classname="tests.core.test_baseline" name="test_baseline_eos_stops_generation" time="0.789" /><testcase classname="tests.core.test_baseline" name="test_baseline_kv_bytes_static_helper" time="0.000" /><testcase classname="tests.core.test_metrics" name="test_cache_kv_bytes_empty_cache" time="1.853" /><testcase classname="tests.core.test_metrics" name="test_cache_kv_bytes_after_prefill" time="2.131" /><testcase classname="tests.core.test_metrics" name="test_cache_token_count_handles_no_layers" time="0.000" /><testcase classname="tests.core.test_metrics" name="test_measure_proposer_weight_bytes" time="2.520" /><testcase classname="tests.core.test_metrics" name="test_report_assembles_all_fields" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_detects_output_divergence" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_handles_zero_output" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_projection_table_shape" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_render_includes_key_fields" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_zero_seq_len_baseline_does_not_divide_by_zero" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_is_serializable" time="0.001" /><testcase classname="tests.core.test_metrics" name="test_report_default_verifier_activation_zero" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_proposer_loads_with_defaults" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_encode_chat_returns_int_list" time="0.007" /><testcase classname="tests.core.test_proposer" name="test_propose_block_returns_unmasked_tokens" time="1.294" /><testcase classname="tests.core.test_proposer" name="test_propose_block_clamps_steps_to_block_size" time="0.581" /><testcase classname="tests.core.test_proposer" name="test_propose_block_updates_running_stats" time="1.244" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_block_size[0]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_block_size[-1]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_block_size[-10]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_num_steps[0]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_num_steps[-1]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_rejects_nonpositive_num_steps[-5]" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_proposer_construct_requires_mask_id" time="2.474" /><testcase classname="tests.core.test_proposer" name="test_proposer_construct_requires_pad_or_eos" time="2.409" /><testcase classname="tests.core.test_proposer" name="test_proposer_stats_dataclass_defaults" time="0.000" /><testcase classname="tests.core.test_proposer" name="test_propose_block_with_smaller_steps_than_block" time="1.062" /><testcase classname="tests.core.test_proposer" name="test_propose_block_underfill_raises" time="1.256" /><testcase classname="tests.core.test_proposer" name="test_encode_chat_rejects_non_list_return" time="0.001" /><testcase classname="tests.core.test_proposer" name="test_encode_chat_rejects_non_int_elements" time="0.001" /><testcase classname="tests.core.test_run_demo_integration" name="test_run_demo_smoke" time="12.880" /><testcase classname="tests.core.test_speculative" name="test_decoder_rejects_nonpositive_block_size[0]" time="1.857" /><testcase classname="tests.core.test_speculative" name="test_decoder_rejects_nonpositive_block_size[-1]" time="1.847" /><testcase classname="tests.core.test_speculative" name="test_decoder_rejects_nonpositive_num_diffusion_steps[0]" time="1.898" /><testcase classname="tests.core.test_speculative" name="test_decoder_rejects_nonpositive_num_diffusion_steps[-1]" time="1.841" /><testcase classname="tests.core.test_speculative" name="test_generate_rejects_nonpositive_max_new_tokens" time="1.941" /><testcase classname="tests.core.test_speculative" name="test_generate_produces_tokens_and_stats" time="2.104" /><testcase classname="tests.core.test_speculative" name="test_generate_with_no_proposed_tokens_returns_zero_acceptance" time="0.000" /><testcase classname="tests.core.test_speculative" name="test_generate_eos_in_accepted_prefix_truncates" time="5.536" /><testcase classname="tests.core.test_speculative" name="test_kv_bytes_static_helper_returns_zero_for_no_cache" time="1.827" /><testcase classname="tests.core.test_speculative" name="test_generate_appends_correction_or_bonus" time="4.929" /><testcase classname="tests.core.test_speculative" name="test_generate_max_new_tokens_short_circuits_after_block" time="4.611" /><testcase classname="tests.core.test_speculative" name="test_generate_correction_token_can_be_eos" time="3.415" /><testcase classname="tests.core.test_speculative" name="test_generate_correction_is_eos_via_rejected_proposal" time="4.182" /><testcase classname="tests.core.test_speculative" name="test_generate_eos_in_middle_of_accepted_block_drops_trailing" time="7.381" /><testcase classname="tests.core.test_verifier" name="test_default_config_loads" time="0.000" /><testcase classname="tests.core.test_verifier" name="test_construction_validates_window_args[-1-8-sink_size must be &gt;= 0]" time="0.000" /><testcase classname="tests.core.test_verifier" name="test_construction_validates_window_args[4-0-window_size must be &gt; 0]" time="0.000" /><testcase classname="tests.core.test_verifier" name="test_construction_validates_window_args[4--3-window_size must be &gt; 0]" time="0.000" /><testcase classname="tests.core.test_verifier" name="test_prefill_rejects_empty" time="0.000" /><testcase classname="tests.core.test_verifier" name="test_prefill_under_budget" time="2.503" /><testcase classname="tests.core.test_verifier" name="test_prefill_over_budget_triggers_trim" time="3.123" /><testcase classname="tests.core.test_verifier" name="test_prefill_zero_sink" time="2.700" /><testcase classname="tests.core.test_verifier" name="test_forward_block_requires_prefill" time="1.905" /><testcase classname="tests.core.test_verifier" name="test_forward_block_rejects_empty" time="2.027" /><testcase classname="tests.core.test_verifier" name="test_forward_block_returns_per_position_logits" time="2.299" /><testcase classname="tests.core.test_verifier" name="test_commit_validates_args" time="2.259" /><testcase classname="tests.core.test_verifier" name="test_commit_full_accept_no_drop" time="2.303" /><testcase classname="tests.core.test_verifier" name="test_commit_partial_accept_drops_tail" time="2.251" /><testcase classname="tests.core.test_verifier" name="test_commit_zero_accept_drops_all" time="2.386" /><testcase classname="tests.core.test_verifier" name="test_commit_post_trims_to_budget" time="2.358" /><testcase classname="tests.core.test_verifier" name="test_append_token_advances_state" time="2.229" /><testcase classname="tests.core.test_verifier" name="test_trim_raises_when_no_cache" time="1.975" /><testcase classname="tests.core.test_verifier" name="test_truncate_tail_raises_when_no_cache" time="1.861" /><testcase classname="tests.core.test_verifier" name="test_truncate_tail_zero_drop_is_noop" time="2.085" /><testcase classname="tests.core.test_verifier" name="test_truncate_tail_overflow_raises" time="2.072" /><testcase classname="tests.core.test_verifier" name="test_trim_detects_layout_violation" time="2.547" /><testcase classname="tests.core.test_verifier" name="test_reset_clears_state" time="2.101" /><testcase classname="tests.core.test_verifier" name="test_trim_skips_layers_with_no_keys" time="2.354" /><testcase classname="tests.core.test_verifier" name="test_truncate_skips_layers_with_no_keys" time="2.385" /><testcase classname="tests.core.test_verifier" name="test_record_peak_kv_handles_null_cache" time="1.832" /><testcase classname="tests.core.test_verifier" name="test_record_peak_kv_handles_layers_with_null_kv" time="2.301" /><testcase classname="tests.core.test_verifier" name="test_record_peak_activation_grows_only" time="2.340" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_proposer_loads" time="2.732" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_proposer_rejects_model_without_backbone" time="2.400" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_proposer_rejects_model_without_lm_head" time="2.621" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_rejects_nonpositive_block_size[0]" time="0.001" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_rejects_nonpositive_block_size[-1]" time="0.000" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_rejects_nonpositive_num_steps[0]" time="0.000" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_rejects_nonpositive_num_steps[-1]" time="0.000" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_clamps_steps_to_block_size" time="0.461" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_returns_unmasked_tokens" time="0.938" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_propose_block_records_activation_peak" time="0.936" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_emits_identical_tokens_to_dense[4-4]" time="2.141" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_emits_identical_tokens_to_dense[8-4]" time="2.413" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_emits_identical_tokens_to_dense[8-3]" time="1.806" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_emits_identical_tokens_to_dense[1-1]" time="0.477" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_emits_identical_tokens_to_dense[16-8]" time="5.887" /><testcase classname="tests.inference_engine.proposer.test_sparse_logits" name="test_sparse_path_activation_peak_smaller_than_dense" time="2.409" /><testcase classname="tests.backends.mlx.test_cache" name="test_construction_defaults" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_construction_validates[-1-8-sink_size must be &gt;= 0]" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_construction_validates[4-0-window_size must be &gt; 0]" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_construction_validates[4--3-window_size must be &gt; 0]" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_first_update_no_trim" time="0.001" /><testcase classname="tests.backends.mlx.test_cache" name="test_subsequent_update_below_budget" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_update_triggers_trim" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_update_validates_kv_dims" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_update_validates_kv_seq_match" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_make_mask_empty_cache_single_token" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_make_mask_empty_cache_multi_token" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_make_mask_after_updates_uses_pre_update_buffer_size" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_make_mask_with_window_size_kwarg" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_trim_empty_cache_is_zero" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_trim_drops_from_end" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_trim_clamped_at_buffer_size" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_trim_zero_is_noop" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_trim_negative_clamped_to_zero" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_size_returns_offset" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_empty" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_nbytes_empty_zero" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_nbytes_positive_after_update" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_is_trimmable" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_state_round_trip" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_make_sink_window_cache_size_matches_model_layers" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_total_kv_bytes_aggregates_layers" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_total_kv_bytes_empty_layers" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_cache_seq_length_picks_first_nonempty" time="0.000" /><testcase classname="tests.backends.mlx.test_cache" name="test_cache_seq_length_all_empty" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_safe_dist_version_missing_returns_none" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_safe_dist_version_present_returns_string" time="0.001" /><testcase classname="tests.backends.mlx.test_env" name="test_non_arm64_returns_unavailable[x86_64]" time="0.020" /><testcase classname="tests.backends.mlx.test_env" name="test_non_arm64_returns_unavailable[i386]" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_non_arm64_returns_unavailable[aarch64]" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_non_arm64_returns_unavailable[ppc64le]" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_arm64_but_mlx_not_installed" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_core_import_raises" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_core_missing_metal_submodule" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_metal_is_available_not_callable" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_metal_is_available_raises" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_metal_is_available_returns_false" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_fully_available" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_mlx_available_without_mlx_lm" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_render_when_available" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_render_when_unavailable" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_require_raises_when_unavailable" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_require_returns_when_available" time="0.000" /><testcase classname="tests.backends.mlx.test_env" name="test_real_probe_is_well_formed" time="0.001" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_loads" time="2.660" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_propose_block_returns_unmasked" time="0.132" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_clamps_steps" time="0.048" /><testcase classname="tests.backends.mlx.test_proposer" name="test_propose_block_rejects_nonpositive_block_size[0]" time="0.000" /><testcase classname="tests.backends.mlx.test_proposer" name="test_propose_block_rejects_nonpositive_block_size[-1]" time="0.000" /><testcase classname="tests.backends.mlx.test_proposer" name="test_propose_block_rejects_nonpositive_num_steps[0]" time="0.000" /><testcase classname="tests.backends.mlx.test_proposer" name="test_propose_block_rejects_nonpositive_num_steps[-1]" time="0.000" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_stats_increment" time="0.113" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_acceptance_by_verifier" time="3.049" /><testcase classname="tests.backends.mlx.test_proposer" name="test_mlx_proposer_sparse_activation_smaller_than_dense_oracle" time="0.135" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_dtype[mx_dtype0-torch_dtype0]" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_dtype[mx_dtype1-torch_dtype1]" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_dtype[mx_dtype2-torch_dtype2]" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_dtype[mx_dtype3-torch_dtype3]" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_values_fp32" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_preserves_values_bf16" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_rejects_non_mx_array" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_torch_to_mx_preserves_dtype[torch_dtype0-mx_dtype0]" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_torch_to_mx_preserves_dtype[torch_dtype1-mx_dtype1]" time="0.000" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_torch_to_mx_preserves_dtype[torch_dtype2-mx_dtype2]" time="0.003" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_torch_to_mx_rejects_non_tensor" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_round_trip_torch_mx_torch_bf16" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_round_trip_mx_torch_mx_fp32" time="0.001" /><testcase classname="tests.backends.mlx.test_torch_bridge" name="test_mx_to_torch_lazy_arrays_get_evaluated" time="0.002" /><testcase classname="tests.backends.mlx.test_verifier" name="test_default_config_loads" time="0.787" /><testcase classname="tests.backends.mlx.test_verifier" name="test_construction_validates_window_args[-1-8-sink_size must be &gt;= 0]" time="0.001" /><testcase classname="tests.backends.mlx.test_verifier" name="test_construction_validates_window_args[4-0-window_size must be &gt; 0]" time="0.001" /><testcase classname="tests.backends.mlx.test_verifier" name="test_construction_rejects_unsupported_dtype" time="1.723" /><testcase classname="tests.backends.mlx.test_verifier" name="test_prefill_rejects_empty" time="0.000" /><testcase classname="tests.backends.mlx.test_verifier" name="test_prefill_under_budget" time="1.405" /><testcase classname="tests.backends.mlx.test_verifier" name="test_prefill_over_budget_triggers_trim" time="0.738" /><testcase classname="tests.backends.mlx.test_verifier" name="test_prefill_zero_sink" time="0.688" /><testcase classname="tests.backends.mlx.test_verifier" name="test_forward_block_requires_prefill" time="0.587" /><testcase classname="tests.backends.mlx.test_verifier" name="test_forward_block_rejects_empty" time="0.636" /><testcase classname="tests.backends.mlx.test_verifier" name="test_forward_block_returns_per_position_logits" time="0.713" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_validates_args" time="0.692" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_full_accept_no_drop" time="0.700" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_partial_accept_drops_tail" time="0.702" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_zero_accept_drops_all" time="0.711" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_post_trims_to_budget" time="0.707" /><testcase classname="tests.backends.mlx.test_verifier" name="test_append_token_advances_state" time="0.702" /><testcase classname="tests.backends.mlx.test_verifier" name="test_cache_buffer_size_zero_when_no_cache" time="0.598" /><testcase classname="tests.backends.mlx.test_verifier" name="test_commit_per_layer_trim_mismatch_raises" time="0.721" /><testcase classname="tests.backends.mlx.test_verifier" name="test_record_peak_kv_handles_null_cache" time="0.626" /><testcase classname="tests.backends.mlx.test_verifier" name="test_record_peak_activation_grows_only" time="0.597" /><testcase classname="tests.backends.mlx.test_verifier" name="test_reset_clears_state" time="0.716" /><testcase classname="tests.backends.mlx.test_verifier" name="test_model_weight_bytes_positive" time="0.001" /><testcase classname="tests.backends.mlx.test_verifier" name="test_dtype_mapping_supported[torch_dtype0-mx_dtype0]" time="0.000" /><testcase classname="tests.backends.mlx.test_verifier" name="test_dtype_mapping_supported[torch_dtype1-mx_dtype1]" time="0.000" /><testcase classname="tests.backends.mlx.test_verifier" name="test_dtype_mapping_supported[torch_dtype2-mx_dtype2]" time="0.000" /><testcase classname="tests.backends.mlx.test_verifier" name="test_dtype_mapping_unsupported_raises" time="0.000" /><testcase classname="tests.backends.mlx.test_verifier" name="test_mlx_argmax_matches_pytorch_baseline" time="3.474" /><testcase classname="tests.backends.mlx.test_verifier" name="test_mlx_long_generation_matches_pytorch_below_budget" time="8.313" /><testcase classname="tests.backends.mlx.test_verifier" name="test_mlx_long_generation_with_trim_matches_pytorch" time="8.572" /></testsuite></testsuites>