Add files using upload-large-folder tool

LTA_openwebtext_dualt/logs/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_gumbel_sde_watch/infer_lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525_step_0020000.log

@@ -0,0 +1,136 @@
+[watch-gumbel] 2026-05-26_02:03:21 infer runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000.pt -> docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000
+[load] runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000.pt
+[ckpt] step=20000
+[sde] generated 2/128
+[sde] generated 4/128
+[sde] generated 6/128
+[sde] generated 8/128
+[sde] generated 10/128
+[sde] generated 12/128
+[sde] generated 14/128
+[sde] generated 16/128
+[sde] generated 18/128
+[sde] generated 20/128
+[sde] generated 22/128
+[sde] generated 24/128
+[sde] generated 26/128
+[sde] generated 28/128
+[sde] generated 30/128
+[sde] generated 32/128
+[sde] generated 34/128
+[sde] generated 36/128
+[sde] generated 38/128
+[sde] generated 40/128
+[sde] generated 42/128
+[sde] generated 44/128
+[sde] generated 46/128
+[sde] generated 48/128
+[sde] generated 50/128
+[sde] generated 52/128
+[sde] generated 54/128
+[sde] generated 56/128
+[sde] generated 58/128
+[sde] generated 60/128
+[sde] generated 62/128
+[sde] generated 64/128
+[sde] generated 66/128
+[sde] generated 68/128
+[sde] generated 70/128
+[sde] generated 72/128
+[sde] generated 74/128
+[sde] generated 76/128
+[sde] generated 78/128
+[sde] generated 80/128
+[sde] generated 82/128
+[sde] generated 84/128
+[sde] generated 86/128
+[sde] generated 88/128
+[sde] generated 90/128
+[sde] generated 92/128
+[sde] generated 94/128
+[sde] generated 96/128
+[sde] generated 98/128
+[sde] generated 100/128
+[sde] generated 102/128
+[sde] generated 104/128
+[sde] generated 106/128
+[sde] generated 108/128
+[sde] generated 110/128
+[sde] generated 112/128
+[sde] generated 114/128
+[sde] generated 116/128
+[sde] generated 118/128
+[sde] generated 120/128
+[sde] generated 122/128
+[sde] generated 124/128
+[sde] generated 126/128
+[sde] generated 128/128
+[score] loading scorer: /e2e-data/evad-tech-vla/wanghan58/models/flowtext_scorers/gpt2-large-standard
+[summary] {
+  "type": "summary",
+  "checkpoint": "runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000.pt",
+  "step": 20000,
+  "decode": {
+    "decode_rule": "dirichlet_resample_sde",
+    "steps": 128,
+    "model_t_mode": "support_t",
+    "mean_mode": "endpoint_only",
+    "anchor_gamma": 1.0,
+    "endpoint_floor": 0.0,
+    "concentration_min": 1.0,
+    "concentration_max": 1024.0,
+    "endpoint_temp": 1.45,
+    "endpoint_temp_start": null,
+    "endpoint_temp_end": null,
+    "endpoint_projection": "gumbel_softmax",
+    "endpoint_top_k": 0,
+    "endpoint_top_p": 0.95,
+    "gumbel_tau_start": 1.0,
+    "gumbel_tau_end": 0.2,
+    "gumbel_noise_scale_start": 1.0,
+    "gumbel_noise_scale_end": 1.0,
+    "ban_special_tokens": false,
+    "banned_endpoint_ids": [],
+    "support_power": 1.0,
+    "semantic_power": 1.0,
+    "noise_init": "dirichlet",
+    "noise_sigma": -1.0,
+    "noise_dirichlet_concentration": 1.0,
+    "sde_resample": "dirichlet",
+    "logistic_normal_sigma_min": 0.18,
+    "logistic_normal_sigma_max": 3.0,
+    "logistic_normal_tau_min": 0.65,
+    "logistic_normal_tau_max": 1.0,
+    "final_from": "blend_0.5",
+    "n_samples": 128,
+    "seed": 20260524
+  },
+  "raw_genppl": {
+    "ppl": 1.3402534767040193,
+    "nll_per_token": 0.2928587577934152,
+    "tokens": 121854,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "stripped_genppl": {
+    "ppl": 1.329592652013063,
+    "nll_per_token": 0.28487261863947294,
+    "tokens": 121693,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "diversity": {
+    "sample_entropy": 0.24651868226929907,
+    "unique_tokens": 262,
+    "token_count": 131072,
+    "distinct_1": 0.0019989013671875,
+    "distinct_2": 0.006392045454545455,
+    "top_token_mass": 0.7519149780273438
+  }
+}
+[done] docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000/sde_steps128_samples128_scored.jsonl
+[watch-gumbel] 2026-05-26_02:10:15 done step_0020000

LTA_openwebtext_dualt/logs/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_gumbel_sde_watch/infer_lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525_step_0050000.log

@@ -0,0 +1,136 @@
+[watch-gumbel] 2026-05-26_07:41:32 infer runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000.pt -> docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000
+[load] runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000.pt
+[ckpt] step=50000
+[sde] generated 2/128
+[sde] generated 4/128
+[sde] generated 6/128
+[sde] generated 8/128
+[sde] generated 10/128
+[sde] generated 12/128
+[sde] generated 14/128
+[sde] generated 16/128
+[sde] generated 18/128
+[sde] generated 20/128
+[sde] generated 22/128
+[sde] generated 24/128
+[sde] generated 26/128
+[sde] generated 28/128
+[sde] generated 30/128
+[sde] generated 32/128
+[sde] generated 34/128
+[sde] generated 36/128
+[sde] generated 38/128
+[sde] generated 40/128
+[sde] generated 42/128
+[sde] generated 44/128
+[sde] generated 46/128
+[sde] generated 48/128
+[sde] generated 50/128
+[sde] generated 52/128
+[sde] generated 54/128
+[sde] generated 56/128
+[sde] generated 58/128
+[sde] generated 60/128
+[sde] generated 62/128
+[sde] generated 64/128
+[sde] generated 66/128
+[sde] generated 68/128
+[sde] generated 70/128
+[sde] generated 72/128
+[sde] generated 74/128
+[sde] generated 76/128
+[sde] generated 78/128
+[sde] generated 80/128
+[sde] generated 82/128
+[sde] generated 84/128
+[sde] generated 86/128
+[sde] generated 88/128
+[sde] generated 90/128
+[sde] generated 92/128
+[sde] generated 94/128
+[sde] generated 96/128
+[sde] generated 98/128
+[sde] generated 100/128
+[sde] generated 102/128
+[sde] generated 104/128
+[sde] generated 106/128
+[sde] generated 108/128
+[sde] generated 110/128
+[sde] generated 112/128
+[sde] generated 114/128
+[sde] generated 116/128
+[sde] generated 118/128
+[sde] generated 120/128
+[sde] generated 122/128
+[sde] generated 124/128
+[sde] generated 126/128
+[sde] generated 128/128
+[score] loading scorer: /e2e-data/evad-tech-vla/wanghan58/models/flowtext_scorers/gpt2-large-standard
+[summary] {
+  "type": "summary",
+  "checkpoint": "runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000.pt",
+  "step": 50000,
+  "decode": {
+    "decode_rule": "dirichlet_resample_sde",
+    "steps": 128,
+    "model_t_mode": "support_t",
+    "mean_mode": "endpoint_only",
+    "anchor_gamma": 1.0,
+    "endpoint_floor": 0.0,
+    "concentration_min": 1.0,
+    "concentration_max": 1024.0,
+    "endpoint_temp": 1.45,
+    "endpoint_temp_start": null,
+    "endpoint_temp_end": null,
+    "endpoint_projection": "gumbel_softmax",
+    "endpoint_top_k": 0,
+    "endpoint_top_p": 0.95,
+    "gumbel_tau_start": 1.0,
+    "gumbel_tau_end": 0.2,
+    "gumbel_noise_scale_start": 1.0,
+    "gumbel_noise_scale_end": 1.0,
+    "ban_special_tokens": false,
+    "banned_endpoint_ids": [],
+    "support_power": 1.0,
+    "semantic_power": 1.0,
+    "noise_init": "dirichlet",
+    "noise_sigma": -1.0,
+    "noise_dirichlet_concentration": 1.0,
+    "sde_resample": "dirichlet",
+    "logistic_normal_sigma_min": 0.18,
+    "logistic_normal_sigma_max": 3.0,
+    "logistic_normal_tau_min": 0.65,
+    "logistic_normal_tau_max": 1.0,
+    "final_from": "blend_0.5",
+    "n_samples": 128,
+    "seed": 20260524
+  },
+  "raw_genppl": {
+    "ppl": 1.0464811518513861,
+    "nll_per_token": 0.045433252088485226,
+    "tokens": 130944,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "stripped_genppl": {
+    "ppl": 1.0318298872771858,
+    "nll_per_token": 0.03133381556240585,
+    "tokens": 130817,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "diversity": {
+    "sample_entropy": 0.007866103229201246,
+    "unique_tokens": 4,
+    "token_count": 131072,
+    "distinct_1": 3.0517578125e-05,
+    "distinct_2": 4.5821114369501466e-05,
+    "top_token_mass": 0.9990081787109375
+  }
+}
+[done] docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000/sde_steps128_samples128_scored.jsonl
+[watch-gumbel] 2026-05-26_07:48:25 done step_0050000

LTA_openwebtext_dualt/logs/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_gumbel_sde_watch/infer_lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525_step_0060000.log

@@ -0,0 +1,136 @@
+[watch-gumbel] 2026-05-26_09:34:28 infer runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000.pt -> docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000
+[load] runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000.pt
+[ckpt] step=60000
+[sde] generated 2/128
+[sde] generated 4/128
+[sde] generated 6/128
+[sde] generated 8/128
+[sde] generated 10/128
+[sde] generated 12/128
+[sde] generated 14/128
+[sde] generated 16/128
+[sde] generated 18/128
+[sde] generated 20/128
+[sde] generated 22/128
+[sde] generated 24/128
+[sde] generated 26/128
+[sde] generated 28/128
+[sde] generated 30/128
+[sde] generated 32/128
+[sde] generated 34/128
+[sde] generated 36/128
+[sde] generated 38/128
+[sde] generated 40/128
+[sde] generated 42/128
+[sde] generated 44/128
+[sde] generated 46/128
+[sde] generated 48/128
+[sde] generated 50/128
+[sde] generated 52/128
+[sde] generated 54/128
+[sde] generated 56/128
+[sde] generated 58/128
+[sde] generated 60/128
+[sde] generated 62/128
+[sde] generated 64/128
+[sde] generated 66/128
+[sde] generated 68/128
+[sde] generated 70/128
+[sde] generated 72/128
+[sde] generated 74/128
+[sde] generated 76/128
+[sde] generated 78/128
+[sde] generated 80/128
+[sde] generated 82/128
+[sde] generated 84/128
+[sde] generated 86/128
+[sde] generated 88/128
+[sde] generated 90/128
+[sde] generated 92/128
+[sde] generated 94/128
+[sde] generated 96/128
+[sde] generated 98/128
+[sde] generated 100/128
+[sde] generated 102/128
+[sde] generated 104/128
+[sde] generated 106/128
+[sde] generated 108/128
+[sde] generated 110/128
+[sde] generated 112/128
+[sde] generated 114/128
+[sde] generated 116/128
+[sde] generated 118/128
+[sde] generated 120/128
+[sde] generated 122/128
+[sde] generated 124/128
+[sde] generated 126/128
+[sde] generated 128/128
+[score] loading scorer: /e2e-data/evad-tech-vla/wanghan58/models/flowtext_scorers/gpt2-large-standard
+[summary] {
+  "type": "summary",
+  "checkpoint": "runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000.pt",
+  "step": 60000,
+  "decode": {
+    "decode_rule": "dirichlet_resample_sde",
+    "steps": 128,
+    "model_t_mode": "support_t",
+    "mean_mode": "endpoint_only",
+    "anchor_gamma": 1.0,
+    "endpoint_floor": 0.0,
+    "concentration_min": 1.0,
+    "concentration_max": 1024.0,
+    "endpoint_temp": 1.45,
+    "endpoint_temp_start": null,
+    "endpoint_temp_end": null,
+    "endpoint_projection": "gumbel_softmax",
+    "endpoint_top_k": 0,
+    "endpoint_top_p": 0.95,
+    "gumbel_tau_start": 1.0,
+    "gumbel_tau_end": 0.2,
+    "gumbel_noise_scale_start": 1.0,
+    "gumbel_noise_scale_end": 1.0,
+    "ban_special_tokens": false,
+    "banned_endpoint_ids": [],
+    "support_power": 1.0,
+    "semantic_power": 1.0,
+    "noise_init": "dirichlet",
+    "noise_sigma": -1.0,
+    "noise_dirichlet_concentration": 1.0,
+    "sde_resample": "dirichlet",
+    "logistic_normal_sigma_min": 0.18,
+    "logistic_normal_sigma_max": 3.0,
+    "logistic_normal_tau_min": 0.65,
+    "logistic_normal_tau_max": 1.0,
+    "final_from": "blend_0.5",
+    "n_samples": 128,
+    "seed": 20260524
+  },
+  "raw_genppl": {
+    "ppl": 1.0477807062851836,
+    "nll_per_token": 0.04667431427519331,
+    "tokens": 126936,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "stripped_genppl": {
+    "ppl": 1.0321466838054298,
+    "nll_per_token": 0.031640792429459864,
+    "tokens": 126797,
+    "kept_samples": 128,
+    "total_samples": 128,
+    "empty_rate": 0.0,
+    "skipped_samples": 0
+  },
+  "diversity": {
+    "sample_entropy": 0.007866103229201246,
+    "unique_tokens": 6,
+    "token_count": 131072,
+    "distinct_1": 4.57763671875e-05,
+    "distinct_2": 7.636852394916911e-05,
+    "top_token_mass": 0.9365692138671875
+  }
+}
+[done] docs/lta_samples/metrics_20260525/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_sde_gumbel_topp0p95_tau1p0_to_0p2_blend_c1_1024_n128/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000/sde_steps128_samples128_scored.jsonl
+[watch-gumbel] 2026-05-26_09:41:20 done step_0060000

LTA_openwebtext_dualt/logs/owt_bert_absrope_time4_C1_to_1024_mask1_sameT_gumbel_sde_watch/processed_lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525_steps128_c1_1024_gumbel_t1p45_n128.txt

@@ -0,0 +1,7 @@
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0010000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0020000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0030000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0040000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0050000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0060000.pt
+runs/lta_owt_bert_absrope_time4_dirichlet_len1024_C1_to_1024_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260525/step_0070000.pt

LTA_openwebtext_dualt/logs/owt_t5elf_absrope_time4_len1025_C1_to_1024_prebos_mask1_sameT_dualline_watch/infer_lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526_step_0010000.log

@@ -0,0 +1,199 @@
+[watch-gumbel] 2026-05-26_13:35:23 infer runs/lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526/step_0010000.pt -> docs/lta_samples/metrics_20260526/owt_t5elf_absrope_time4_len1025_C1_to_1024_prebos_mask1_sameT_dualline_dirres_c1_1024_n128/lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526/step_0010000
+[decode] max_len=1025 generated=1/128
+[decode] max_len=1025 generated=2/128
+[decode] max_len=1025 generated=3/128
+[decode] max_len=1025 generated=4/128
+[decode] max_len=1025 generated=5/128
+[decode] max_len=1025 generated=6/128
+[decode] max_len=1025 generated=7/128
+[decode] max_len=1025 generated=8/128
+[decode] max_len=1025 generated=9/128
+[decode] max_len=1025 generated=10/128
+[decode] max_len=1025 generated=11/128
+[decode] max_len=1025 generated=12/128
+[decode] max_len=1025 generated=13/128
+[decode] max_len=1025 generated=14/128
+[decode] max_len=1025 generated=15/128
+[decode] max_len=1025 generated=16/128
+[decode] max_len=1025 generated=17/128
+[decode] max_len=1025 generated=18/128
+[decode] max_len=1025 generated=19/128
+[decode] max_len=1025 generated=20/128
+[decode] max_len=1025 generated=21/128
+[decode] max_len=1025 generated=22/128
+[decode] max_len=1025 generated=23/128
+[decode] max_len=1025 generated=24/128
+[decode] max_len=1025 generated=25/128
+[decode] max_len=1025 generated=26/128
+[decode] max_len=1025 generated=27/128
+[decode] max_len=1025 generated=28/128
+[decode] max_len=1025 generated=29/128
+[decode] max_len=1025 generated=30/128
+[decode] max_len=1025 generated=31/128
+[decode] max_len=1025 generated=32/128
+[decode] max_len=1025 generated=33/128
+[decode] max_len=1025 generated=34/128
+[decode] max_len=1025 generated=35/128
+[decode] max_len=1025 generated=36/128
+[decode] max_len=1025 generated=37/128
+[decode] max_len=1025 generated=38/128
+[decode] max_len=1025 generated=39/128
+[decode] max_len=1025 generated=40/128
+[decode] max_len=1025 generated=41/128
+[decode] max_len=1025 generated=42/128
+[decode] max_len=1025 generated=43/128
+[decode] max_len=1025 generated=44/128
+[decode] max_len=1025 generated=45/128
+[decode] max_len=1025 generated=46/128
+[decode] max_len=1025 generated=47/128
+[decode] max_len=1025 generated=48/128
+[decode] max_len=1025 generated=49/128
+[decode] max_len=1025 generated=50/128
+[decode] max_len=1025 generated=51/128
+[decode] max_len=1025 generated=52/128
+[decode] max_len=1025 generated=53/128
+[decode] max_len=1025 generated=54/128
+[decode] max_len=1025 generated=55/128
+[decode] max_len=1025 generated=56/128
+[decode] max_len=1025 generated=57/128
+[decode] max_len=1025 generated=58/128
+[decode] max_len=1025 generated=59/128
+[decode] max_len=1025 generated=60/128
+[decode] max_len=1025 generated=61/128
+[decode] max_len=1025 generated=62/128
+[decode] max_len=1025 generated=63/128
+[decode] max_len=1025 generated=64/128
+[decode] max_len=1025 generated=65/128
+[decode] max_len=1025 generated=66/128
+[decode] max_len=1025 generated=67/128
+[decode] max_len=1025 generated=68/128
+[decode] max_len=1025 generated=69/128
+[decode] max_len=1025 generated=70/128
+[decode] max_len=1025 generated=71/128
+[decode] max_len=1025 generated=72/128
+[decode] max_len=1025 generated=73/128
+[decode] max_len=1025 generated=74/128
+[decode] max_len=1025 generated=75/128
+[decode] max_len=1025 generated=76/128
+[decode] max_len=1025 generated=77/128
+[decode] max_len=1025 generated=78/128
+[decode] max_len=1025 generated=79/128
+[decode] max_len=1025 generated=80/128
+[decode] max_len=1025 generated=81/128
+[decode] max_len=1025 generated=82/128
+[decode] max_len=1025 generated=83/128
+[decode] max_len=1025 generated=84/128
+[decode] max_len=1025 generated=85/128
+[decode] max_len=1025 generated=86/128
+[decode] max_len=1025 generated=87/128
+[decode] max_len=1025 generated=88/128
+[decode] max_len=1025 generated=89/128
+[decode] max_len=1025 generated=90/128
+[decode] max_len=1025 generated=91/128
+[decode] max_len=1025 generated=92/128
+[decode] max_len=1025 generated=93/128
+[decode] max_len=1025 generated=94/128
+[decode] max_len=1025 generated=95/128
+[decode] max_len=1025 generated=96/128
+[decode] max_len=1025 generated=97/128
+[decode] max_len=1025 generated=98/128
+[decode] max_len=1025 generated=99/128
+[decode] max_len=1025 generated=100/128
+[decode] max_len=1025 generated=101/128
+[decode] max_len=1025 generated=102/128
+[decode] max_len=1025 generated=103/128
+[decode] max_len=1025 generated=104/128
+[decode] max_len=1025 generated=105/128
+[decode] max_len=1025 generated=106/128
+[decode] max_len=1025 generated=107/128
+[decode] max_len=1025 generated=108/128
+[decode] max_len=1025 generated=109/128
+[decode] max_len=1025 generated=110/128
+[decode] max_len=1025 generated=111/128
+[decode] max_len=1025 generated=112/128
+[decode] max_len=1025 generated=113/128
+[decode] max_len=1025 generated=114/128
+[decode] max_len=1025 generated=115/128
+[decode] max_len=1025 generated=116/128
+[decode] max_len=1025 generated=117/128
+[decode] max_len=1025 generated=118/128
+[decode] max_len=1025 generated=119/128
+[decode] max_len=1025 generated=120/128
+[decode] max_len=1025 generated=121/128
+[decode] max_len=1025 generated=122/128
+[decode] max_len=1025 generated=123/128
+[decode] max_len=1025 generated=124/128
+[decode] max_len=1025 generated=125/128
+[decode] max_len=1025 generated=126/128
+[decode] max_len=1025 generated=127/128
+[decode] max_len=1025 generated=128/128
+[
+  {
+    "checkpoint": "runs/lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526/step_0010000.pt",
+    "ckpt_step": 10000,
+    "max_len": 1025,
+    "decode_rule": "dual_line_resample",
+    "support_power": 1.0,
+    "semantic_power": 1.5,
+    "steps": 128,
+    "c_min": 1.0,
+    "c_max": 1024.0,
+    "anchor_mode": "state",
+    "model_t_mode": "flow",
+    "time_schedule": "uniform",
+    "time_logit_mean": -1.5,
+    "time_logit_std": 0.8,
+    "time_power": 2.0,
+    "input_noise_scale": 0.0,
+    "input_noise_until": 1.0,
+    "input_noise_dirichlet_concentration": 1.0,
+    "endpoint_softening": "none",
+    "endpoint_soft_power": 2.0,
+    "endpoint_soft_min_conf": 0.0,
+    "endpoint_soft_max_conf": 1.0,
+    "soft_target_decode_mode": "off",
+    "soft_target_power": 1.0,
+    "soft_target_min_conf": 0.0,
+    "soft_target_max_conf": 1.0,
+    "soft_target_debias_start": 0.7,
+    "final_from": "blend",
+    "final_decode": "argmax",
+    "final_sample_temp": 1.0,
+    "final_top_k": 0,
+    "final_top_p": 1.0,
+    "commit_mode": "off",
+    "commit_conf_threshold": 0.0,
+    "commit_margin_threshold": 0.0,
+    "commit_start": 0.0,
+    "commit_min_ratio": 0.0,
+    "commit_max_ratio": 1.0,
+    "commit_power": 2.0,
+    "commit_freq_max_frac": 0.08,
+    "early_temp": 2.8,
+    "late_temp": 1.45,
+    "temp_end": 0.55,
+    "temp_power": 1.5,
+    "pos_extend": "repeat",
+    "fixed_first_token_id": null,
+    "fixed_first_token_text": "",
+    "fixed_first_initial_argmax": false,
+    "use_ema": false,
+    "n_samples": 128,
+    "sample_entropy": 1.5424398482114823,
+    "unique_tokens": 808,
+    "token_count": 131200,
+    "distinct_1": 0.006158536585365854,
+    "distinct_2": 0.04183197021484375,
+    "top_token_mass": 0.1704496951219512,
+    "texts_preview": [
+      "</s> , , 2016 ( e the year of the s- the part of the s, The volume of the quarter of the year, observed in the year of the s ., c,e the 2012 s ) The distribution of the total reduction of the quarter, the s- the figures of the quarter- the s- the s of the s . s,, the year, in the top- the fourth quarter of the total of the year of 2016,- the volume of the s,- the period, ,- the increase of the year- s , on the fourth quarter,- the the quarter of the total of the year of the year, in the s . The year of the 2015 , the year of the year of the period, the year of 2016,- the top quarter , 2013 . The increase of the s ,- the 2015, from the year of the quarter,- the year of 2016,- a large part of the 2015 ,- the- year , in the s, 2012 , as well as the cause of the s,- the year, from the fourth year, in the the--,, in- the, ,e,, e 2009,,ee, 2012,,,e, year , thee52 year ,,,eee 2015,,, ,,,e 2012 -,e thee part 2015-- 2012,,, the year year year,,, ,, ,,888 the- year year, - the year year,, , ,,,, , ,203888 the year year year the 2008,,, the year year year ,,,e year year the 2012,,, ,,, the year yeard the year year ,,, part , 2009,, , part, 2015,,, thee year , 2012, thee, year, 2012,,,, ,,, the year of, 2012 ,,, ,, ,,,e thee year 2008 2012,, , ,,,, thee year,, year ,,,s i...,,, ,,e the year- the- ,,e, the 2008 year,, ,, 2012 year,,, the year ,,, ,, ,- ,,, the year, year,, 2012 , ,,,, ,,,, the, 2008 , , ,-,, ,, , , the 2009 , the,e year, ,,,,,,,,,,, part,, 2008 2012 -,,- ,- ,,, ,-,, ,,, year, the the quarter year , -,,,,, year232, the 2012, , the year, year,,, ,, ,232, the year year year 2008,, year year ,,,,,e the year year the year, 2009,,, 2008 , year, the year- ,,, , the e quarter-,, year, the ,,, the, the year the 2006 ,- year, year-,, 2015, the 2012- year year -, - year the ,-,, , , the part, the ,,, -,,,- ,, ,,,, ,, , , the 2009e,,, the-, 2009, ,,,,--- the - period, year, ,a , , -,s 2009, from ,,, ,, , e isi. ., -,-,- 2012 2009,,,e part, the 2006, ,,, , , ,e, year -,, 2012e,- , ,, , ,,,, ,e ,, e,e,, , ,, e ,,, , ,- e, ,e ,,,eee the 2006,-,- - year,, 2009-, 2012- ,, ,,,e year,,?-- ,- - 2015,,,",
+      "</s> “<unk>, you <unk>,’ or , you have?,<unk>,’, you . ’’<unk>,’ you know it? . . ., . <unk>,’, you . . . . ., in a ,’ you know that, and you must forget it... <unk>,’ pym , you know it. ... you don’t put it, you with a , said “ , I have a , is it ?” “?<unk>,’, you , ? <unk> ” “<unk> ? What? you know you is be’ if <unk> wasn’t that . . . <unk> h,’ if you want , you could do with, I’d be with <unk>.’, you said , you get discussion of the experiences, and what you get that? <unk> , you may have that,’ you know, you get on a . . ., a?’ , you?, <unk>, . , you s<unk> if you get this report,’<unk> you <unk> <unk>’, you<unk> <unk><unk>. ..,’ you, , <unk>, you <unk> <unk><unk>’, you <unk>, <unk><unk><unk><unk><unk><unk> ?<unk><unk>?<unk><unk><unk> ? <unk>,<unk>’.,, you. ’, you<unk> <unk><unk>, you<unk>?<unk> <unk>,’ at you ’,.’, you,,<unk> <unk> <unk>. you’. know’ that you,<unk>’’ said’’e?<unk><unk> ’,, you’ you,’ ,<unk>’’’’’ you’, <unk>. , said, said’’<unk><unk> <unk> <unk> <unk>’ <unk><unk> <unk><unk><unk>,’ youd, ’, .<unk> ,’ you<unk>,<unk>, ?’ ,<unk>, ’’<unk>’ .. ’. you <unk><unk><unk>’’’,’’?’’,’’’172,,’’’199 , ’’,’d’196196196172’,d ’146’888,,’’196165196172,1/2,’ D’’’’.’7771/2-172, ’’<unk>172d,172-188,’ ’’ <unk>?<unk><unk>,<unk><unk>? <unk> <unk><unk> <unk>?<unk> ?<unk>186<unk>,,<unk> <unk>,<unk>’<unk>?<unk>? ?<unk><unk><unk> .’ ’’?d’186,211<unk>,<unk>,<unk> , ’, <unk>?<unk><unk><unk>? <unk> ?<unk> ?<unk><unk>, <unk>. 186.,<unk>., ,<unk>.777,,480’<unk> ,1721/2,1/2 .<unk> 186,,,150<unk>?196<unk>172232?,, ., ., .’172ACS-196172,180,<unk>s’ ... <unk> <unk>152<unk>1/2,430D ,430-D<unk>,<unk>,<unk>’<unk>,’’D’’187 -172172s1/2 .,. <unk>,<unk>,186<unk>333<unk>? ?.<unk><unk> ,186<unk> .’.’’’’<unk>,’’’’’<unk>’, , ,,96<unk>430<unk> , , <unk><unk> ,<unk> <unk> <unk>’,’’’ ’<unk>’ <unk>’’’’’’’’’ ’,<unk>’ ’<unk>’ ’’,<unk>,’,<unk> <unk>?<unk> <unk>’,,’’’<unk> ,.<unk><unk>., ?, <unk><unk> <unk><unk><unk>?, <unk> <unk>’<unk>, at , ,.<unk><unk> .. .<unk>.’ . , .’,,?, ?’<unk>, .’’ ’ ., ’?,,’<unk><unk>, .. <unk> .’,<unk><unk>,<unk>’,,?<unk>’ ’<unk>?<unk>?’’’’’<unk>?<unk><unk>’,’’’m? <unk> <unk>, <unk>, ,, ., .’, .’ ,,<unk>.? ,,",
+      "</s> <unk> SSD ' <unk> SSD ' ' ' <unk> SSD ' ' ' ' ' ' ' ' ' ' ' ' ' <unk> SSD ' ' ' <unk> SSD '' <unk> SSD ' ' <unk> SSD ' <unk> SSD ' ' ' <unk> SSD ' <unk> ' ' <unk> SSD ' ' ' ' ' ' ' ' ' <unk> SSD ' ' ' ' <unk> ' ' <unk> SSD ' ' <unk> SSD ' <unk> SSD ' <unk> <unk> <unk> SSD ' <unk> <unk> SSD ' <unk> SSD ' <unk> SSD <unk> SSD ' ' ' ' <unk> ' <unk> <unk> <unk> <unk> <unk> SSD <unk> SSD ' ' <unk> SSD ' ' <unk> SSD ' <unk> <unk> SSD <unk> SSD ' <unk> ' <unk> <unk> <unk> ' ' ' ' <unk> SSD ' ' <unk> <unk> <unk> 'tt;' SSD with00x0xxx0xx00x0xxxxxxxxxxxxx0xxxxxxxxxxxxxxxxxxxxx0xx0xxxxxxxxxxx0xx00xxxx0xxxxxxxxxxxxx0xxxxxxxxxxxxx0xxx000xxxxxxx0xxxxxxxxxxxxxxxxxxxxxx00xx0x0xxxx0x0xx0xx0xxxxxxxx000x0xx0xxx000xx00000xx0xx0xx0x00xxxxxx0xxxxx0x0xxxxx000x00xx0x0xx0x0xx0x0000xxx00x000xx00x00x0xx00000x000xxxxx0xxx000x0x0xx0x00xx0000x0xxx000x00000000000000x00000000000x0x00x000000000x0x00x0000x00xx00x000x000000;0;;;;;x00xx0000000x00000x00xxx00xxxx0x000x0x000000x0x0x0000xx000xx000000xxx0x0x0x0xxx000000x0x00x0xxx000x00000000xx00000000x0x000x00x000xxxxx000x0x000x00xx00x000x0x0000xxx00000xx0x0x00000x0000x0xxxx00xxxxxx000000x0x00x0x000xx0xxxx00xxxxxx00xxxx0xx0xxxx0x00xxxxxx00xxxxxxx00xxx0000xx0x0x00xx0xxxxxxxxxxxx0x00xxx;;;;;;;;;0;x0x0xx;x;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;",
+      "</s> The <unk> is a <unk>- g , a t <unk> is a -e , a sc <unk> / .,t know what is a dy -e , a , ad ,e , ad ,t getz what is exactly what is om, what, dv ( <unk>): I. what. . d , m -e , a , ,? dd , s d , ? om's ad , what's do o <unk> , daw, e, -e , l<unk> oc d) <unk>e <unk>a -e . a s s d , a i.e. dm, dt<unk> o- pn ,e ms- , a- dl gyzn , a s- <unk> ew -<unk>, arh pn-)<unk>in <unk> , <unk>, <unk>,<unk><unk> /o<unk> ( ()n-d-n, d- den - , d'g, ,'' ( a)<unk> 's,', ''. ''d', is,',''' (' -- dd ,-)--' is is what' ''' '' , ' in, \"' ( (s )n e e,<unk><unk>,d),,ad- ,r . '' isa ' - t is '' - -c---, d, ,- <unk> <unk>ooc ,<unk> <unk>, w',) :,' a, , ,' - <unk>i <unk><unk><unk>, <unk> <unk><unk>, <unk> e :d'''s <unk>', s' g c -w <unk> <unk> e<unk> i t 'l s eses -d nte im d, <unk> <unk><unk><unk><unk>r. <unk>ele.,, ) ...i p<unk>o .b.p -nsa d a ad d' <unk><unk>,e h <unk> <unk> <unk>d'' , ,'-'''','?','<unk>',' ''. ,' ' is a<unk> (d'').,<unk>. ,c (p' a,r )<unk><unk><unk> what<unk>'s? it s - what, have that’, have'''',' ' ? ?!?s (- )r (/, .cl<unk> nci, (<unk> <unk>-e ). .g sbr ()-<unk>r. .- <unk>r,r. a-<unk>g <unk><unk><unk><unk> , n c<unk><unk> <unk><unk>/ <unk>: aa s, . ?i a-d <unk> c<unk>b , .<unk><unk><unk><unk><unk><unk> en .,bm,r<unk>'<unk>, <unk> ,w ,<unk> d .d ,<unk>,<unk>a, o<unk>w,/<unk>/<unk> s o-<unk> <unk>d) asa <unk> , ,ens .nsand, <unk>.. .o<unk>w.,,- o e, (.<unk>.w ,..eo v n)<unk>-- (y<unk><unk> ,az' <unk>,w) ,-<unk> <unk><unk><unk>r ( al <unk> a<unk> ,m<unk>r<unk> -e-)-a b ,a,<unk> e .<unk> .<unk><unk> -., )v <unk><unk><unk>."
+    ],
+    "gen_ppl": 5.966201276828079,
+    "gen_nll": 1.7861104228396043,
+    "gen_tokens": 95099
+  }
+]
+[watch-gumbel] 2026-05-26_13:41:20 done step_0010000

LTA_openwebtext_dualt/logs/owt_t5elf_absrope_time4_len1025_C1_to_1024_prebos_mask1_sameT_dualline_watch/processed_lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526_steps128_c1_1024_gumbel_t1p45_n128.txt

@@ -0,0 +1 @@
+runs/lta_owt_t5elf_absrope_time4_dirichlet_len1025_C1_to_1024_prebos_mask1_sameT_gbs512_b32_8gpu_1m_save10k_20260526/step_0010000.pt

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/array_constructors.py

@@ -0,0 +1,137 @@
+import sys
+from typing import Any
+import numpy as np
+
+
+class Index:
+    def __index__(self) -> int:
+        return 0
+
+
+class SubClass(np.ndarray):
+    pass
+
+
+def func(i: int, j: int, **kwargs: Any) -> SubClass:
+    return B
+
+
+i8 = np.int64(1)
+
+A = np.array([1])
+B = A.view(SubClass).copy()
+B_stack = np.array([[1], [1]]).view(SubClass)
+C = [1]
+
+np.ndarray(Index())
+np.ndarray([Index()])
+
+np.array(1, dtype=float)
+np.array(1, copy=False)
+np.array(1, order='F')
+np.array(1, order=None)
+np.array(1, subok=True)
+np.array(1, ndmin=3)
+np.array(1, str, copy=True, order='C', subok=False, ndmin=2)
+
+np.asarray(A)
+np.asarray(B)
+np.asarray(C)
+
+np.asanyarray(A)
+np.asanyarray(B)
+np.asanyarray(B, dtype=int)
+np.asanyarray(C)
+
+np.ascontiguousarray(A)
+np.ascontiguousarray(B)
+np.ascontiguousarray(C)
+
+np.asfortranarray(A)
+np.asfortranarray(B)
+np.asfortranarray(C)
+
+np.require(A)
+np.require(B)
+np.require(B, dtype=int)
+np.require(B, requirements=None)
+np.require(B, requirements="E")
+np.require(B, requirements=["ENSUREARRAY"])
+np.require(B, requirements={"F", "E"})
+np.require(B, requirements=["C", "OWNDATA"])
+np.require(B, requirements="W")
+np.require(B, requirements="A")
+np.require(C)
+
+np.linspace(0, 2)
+np.linspace(0.5, [0, 1, 2])
+np.linspace([0, 1, 2], 3)
+np.linspace(0j, 2)
+np.linspace(0, 2, num=10)
+np.linspace(0, 2, endpoint=True)
+np.linspace(0, 2, retstep=True)
+np.linspace(0j, 2j, retstep=True)
+np.linspace(0, 2, dtype=bool)
+np.linspace([0, 1], [2, 3], axis=Index())
+
+np.logspace(0, 2, base=2)
+np.logspace(0, 2, base=2)
+np.logspace(0, 2, base=[1j, 2j], num=2)
+
+np.geomspace(1, 2)
+
+np.zeros_like(A)
+np.zeros_like(C)
+np.zeros_like(B)
+np.zeros_like(B, dtype=np.int64)
+
+np.ones_like(A)
+np.ones_like(C)
+np.ones_like(B)
+np.ones_like(B, dtype=np.int64)
+
+np.empty_like(A)
+np.empty_like(C)
+np.empty_like(B)
+np.empty_like(B, dtype=np.int64)
+
+np.full_like(A, i8)
+np.full_like(C, i8)
+np.full_like(B, i8)
+np.full_like(B, i8, dtype=np.int64)
+
+np.ones(1)
+np.ones([1, 1, 1])
+
+np.full(1, i8)
+np.full([1, 1, 1], i8)
+
+np.indices([1, 2, 3])
+np.indices([1, 2, 3], sparse=True)
+
+np.fromfunction(func, (3, 5))
+
+np.identity(10)
+
+np.atleast_1d(C)
+np.atleast_1d(A)
+np.atleast_1d(C, C)
+np.atleast_1d(C, A)
+np.atleast_1d(A, A)
+
+np.atleast_2d(C)
+
+np.atleast_3d(C)
+
+np.vstack([C, C])
+np.vstack([C, A])
+np.vstack([A, A])
+
+np.hstack([C, C])
+
+np.stack([C, C])
+np.stack([C, C], axis=0)
+np.stack([C, C], out=B_stack)
+
+np.block([[C, C], [C, C]])
+np.block(A)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/array_like.py

@@ -0,0 +1,41 @@
+from __future__ import annotations
+
+from typing import Any
+
+import numpy as np
+from numpy._typing import ArrayLike, _SupportsArray
+
+x1: ArrayLike = True
+x2: ArrayLike = 5
+x3: ArrayLike = 1.0
+x4: ArrayLike = 1 + 1j
+x5: ArrayLike = np.int8(1)
+x6: ArrayLike = np.float64(1)
+x7: ArrayLike = np.complex128(1)
+x8: ArrayLike = np.array([1, 2, 3])
+x9: ArrayLike = [1, 2, 3]
+x10: ArrayLike = (1, 2, 3)
+x11: ArrayLike = "foo"
+x12: ArrayLike = memoryview(b'foo')
+
+
+class A:
+    def __array__(self, dtype: None | np.dtype[Any] = None) -> np.ndarray:
+        return np.array([1, 2, 3])
+
+
+x13: ArrayLike = A()
+
+scalar: _SupportsArray = np.int64(1)
+scalar.__array__()
+array: _SupportsArray = np.array(1)
+array.__array__()
+
+a: _SupportsArray = A()
+a.__array__()
+a.__array__()
+
+# Escape hatch for when you mean to make something like an object
+# array.
+object_array_scalar: Any = (i for i in range(10))
+np.array(object_array_scalar)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/arrayterator.py

@@ -0,0 +1,27 @@
+
+from __future__ import annotations
+
+from typing import Any
+import numpy as np
+
+AR_i8: np.ndarray[Any, np.dtype[np.int_]] = np.arange(10)
+ar_iter = np.lib.Arrayterator(AR_i8)
+
+ar_iter.var
+ar_iter.buf_size
+ar_iter.start
+ar_iter.stop
+ar_iter.step
+ar_iter.shape
+ar_iter.flat
+
+ar_iter.__array__()
+
+for i in ar_iter:
+    pass
+
+ar_iter[0]
+ar_iter[...]
+ar_iter[:]
+ar_iter[0, 0, 0]
+ar_iter[..., 0, :]

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/comparisons.py

@@ -0,0 +1,301 @@
+from __future__ import annotations
+
+from typing import Any
+import numpy as np
+
+c16 = np.complex128()
+f8 = np.float64()
+i8 = np.int64()
+u8 = np.uint64()
+
+c8 = np.complex64()
+f4 = np.float32()
+i4 = np.int32()
+u4 = np.uint32()
+
+dt = np.datetime64(0, "D")
+td = np.timedelta64(0, "D")
+
+b_ = np.bool_()
+
+b = bool()
+c = complex()
+f = float()
+i = int()
+
+SEQ = (0, 1, 2, 3, 4)
+
+AR_b: np.ndarray[Any, np.dtype[np.bool_]] = np.array([True])
+AR_u: np.ndarray[Any, np.dtype[np.uint32]] = np.array([1], dtype=np.uint32)
+AR_i: np.ndarray[Any, np.dtype[np.int_]] = np.array([1])
+AR_f: np.ndarray[Any, np.dtype[np.float_]] = np.array([1.0])
+AR_c: np.ndarray[Any, np.dtype[np.complex_]] = np.array([1.0j])
+AR_m: np.ndarray[Any, np.dtype[np.timedelta64]] = np.array([np.timedelta64("1")])
+AR_M: np.ndarray[Any, np.dtype[np.datetime64]] = np.array([np.datetime64("1")])
+AR_O: np.ndarray[Any, np.dtype[np.object_]] = np.array([1], dtype=object)
+
+# Arrays
+
+AR_b > AR_b
+AR_b > AR_u
+AR_b > AR_i
+AR_b > AR_f
+AR_b > AR_c
+
+AR_u > AR_b
+AR_u > AR_u
+AR_u > AR_i
+AR_u > AR_f
+AR_u > AR_c
+
+AR_i > AR_b
+AR_i > AR_u
+AR_i > AR_i
+AR_i > AR_f
+AR_i > AR_c
+
+AR_f > AR_b
+AR_f > AR_u
+AR_f > AR_i
+AR_f > AR_f
+AR_f > AR_c
+
+AR_c > AR_b
+AR_c > AR_u
+AR_c > AR_i
+AR_c > AR_f
+AR_c > AR_c
+
+AR_m > AR_b
+AR_m > AR_u
+AR_m > AR_i
+AR_b > AR_m
+AR_u > AR_m
+AR_i > AR_m
+
+AR_M > AR_M
+
+AR_O > AR_O
+1 > AR_O
+AR_O > 1
+
+# Time structures
+
+dt > dt
+
+td > td
+td > i
+td > i4
+td > i8
+td > AR_i
+td > SEQ
+
+# boolean
+
+b_ > b
+b_ > b_
+b_ > i
+b_ > i8
+b_ > i4
+b_ > u8
+b_ > u4
+b_ > f
+b_ > f8
+b_ > f4
+b_ > c
+b_ > c16
+b_ > c8
+b_ > AR_i
+b_ > SEQ
+
+# Complex
+
+c16 > c16
+c16 > f8
+c16 > i8
+c16 > c8
+c16 > f4
+c16 > i4
+c16 > b_
+c16 > b
+c16 > c
+c16 > f
+c16 > i
+c16 > AR_i
+c16 > SEQ
+
+c16 > c16
+f8 > c16
+i8 > c16
+c8 > c16
+f4 > c16
+i4 > c16
+b_ > c16
+b > c16
+c > c16
+f > c16
+i > c16
+AR_i > c16
+SEQ > c16
+
+c8 > c16
+c8 > f8
+c8 > i8
+c8 > c8
+c8 > f4
+c8 > i4
+c8 > b_
+c8 > b
+c8 > c
+c8 > f
+c8 > i
+c8 > AR_i
+c8 > SEQ
+
+c16 > c8
+f8 > c8
+i8 > c8
+c8 > c8
+f4 > c8
+i4 > c8
+b_ > c8
+b > c8
+c > c8
+f > c8
+i > c8
+AR_i > c8
+SEQ > c8
+
+# Float
+
+f8 > f8
+f8 > i8
+f8 > f4
+f8 > i4
+f8 > b_
+f8 > b
+f8 > c
+f8 > f
+f8 > i
+f8 > AR_i
+f8 > SEQ
+
+f8 > f8
+i8 > f8
+f4 > f8
+i4 > f8
+b_ > f8
+b > f8
+c > f8
+f > f8
+i > f8
+AR_i > f8
+SEQ > f8
+
+f4 > f8
+f4 > i8
+f4 > f4
+f4 > i4
+f4 > b_
+f4 > b
+f4 > c
+f4 > f
+f4 > i
+f4 > AR_i
+f4 > SEQ
+
+f8 > f4
+i8 > f4
+f4 > f4
+i4 > f4
+b_ > f4
+b > f4
+c > f4
+f > f4
+i > f4
+AR_i > f4
+SEQ > f4
+
+# Int
+
+i8 > i8
+i8 > u8
+i8 > i4
+i8 > u4
+i8 > b_
+i8 > b
+i8 > c
+i8 > f
+i8 > i
+i8 > AR_i
+i8 > SEQ
+
+u8 > u8
+u8 > i4
+u8 > u4
+u8 > b_
+u8 > b
+u8 > c
+u8 > f
+u8 > i
+u8 > AR_i
+u8 > SEQ
+
+i8 > i8
+u8 > i8
+i4 > i8
+u4 > i8
+b_ > i8
+b > i8
+c > i8
+f > i8
+i > i8
+AR_i > i8
+SEQ > i8
+
+u8 > u8
+i4 > u8
+u4 > u8
+b_ > u8
+b > u8
+c > u8
+f > u8
+i > u8
+AR_i > u8
+SEQ > u8
+
+i4 > i8
+i4 > i4
+i4 > i
+i4 > b_
+i4 > b
+i4 > AR_i
+i4 > SEQ
+
+u4 > i8
+u4 > i4
+u4 > u8
+u4 > u4
+u4 > i
+u4 > b_
+u4 > b
+u4 > AR_i
+u4 > SEQ
+
+i8 > i4
+i4 > i4
+i > i4
+b_ > i4
+b > i4
+AR_i > i4
+SEQ > i4
+
+i8 > u4
+i4 > u4
+u8 > u4
+u4 > u4
+b_ > u4
+b > u4
+i > u4
+AR_i > u4
+SEQ > u4

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/flatiter.py

@@ -0,0 +1,16 @@
+import numpy as np
+
+a = np.empty((2, 2)).flat
+
+a.base
+a.copy()
+a.coords
+a.index
+iter(a)
+next(a)
+a[0]
+a[[0, 1, 2]]
+a[...]
+a[:]
+a.__array__()
+a.__array__(np.dtype(np.float64))

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/fromnumeric.py

@@ -0,0 +1,260 @@
+"""Tests for :mod:`numpy.core.fromnumeric`."""
+
+import numpy as np
+
+A = np.array(True, ndmin=2, dtype=bool)
+B = np.array(1.0, ndmin=2, dtype=np.float32)
+A.setflags(write=False)
+B.setflags(write=False)
+
+a = np.bool_(True)
+b = np.float32(1.0)
+c = 1.0
+d = np.array(1.0, dtype=np.float32)  # writeable
+
+np.take(a, 0)
+np.take(b, 0)
+np.take(c, 0)
+np.take(A, 0)
+np.take(B, 0)
+np.take(A, [0])
+np.take(B, [0])
+
+np.reshape(a, 1)
+np.reshape(b, 1)
+np.reshape(c, 1)
+np.reshape(A, 1)
+np.reshape(B, 1)
+
+np.choose(a, [True, True])
+np.choose(A, [1.0, 1.0])
+
+np.repeat(a, 1)
+np.repeat(b, 1)
+np.repeat(c, 1)
+np.repeat(A, 1)
+np.repeat(B, 1)
+
+np.swapaxes(A, 0, 0)
+np.swapaxes(B, 0, 0)
+
+np.transpose(a)
+np.transpose(b)
+np.transpose(c)
+np.transpose(A)
+np.transpose(B)
+
+np.partition(a, 0, axis=None)
+np.partition(b, 0, axis=None)
+np.partition(c, 0, axis=None)
+np.partition(A, 0)
+np.partition(B, 0)
+
+np.argpartition(a, 0)
+np.argpartition(b, 0)
+np.argpartition(c, 0)
+np.argpartition(A, 0)
+np.argpartition(B, 0)
+
+np.sort(A, 0)
+np.sort(B, 0)
+
+np.argsort(A, 0)
+np.argsort(B, 0)
+
+np.argmax(A)
+np.argmax(B)
+np.argmax(A, axis=0)
+np.argmax(B, axis=0)
+
+np.argmin(A)
+np.argmin(B)
+np.argmin(A, axis=0)
+np.argmin(B, axis=0)
+
+np.searchsorted(A[0], 0)
+np.searchsorted(B[0], 0)
+np.searchsorted(A[0], [0])
+np.searchsorted(B[0], [0])
+
+np.resize(a, (5, 5))
+np.resize(b, (5, 5))
+np.resize(c, (5, 5))
+np.resize(A, (5, 5))
+np.resize(B, (5, 5))
+
+np.squeeze(a)
+np.squeeze(b)
+np.squeeze(c)
+np.squeeze(A)
+np.squeeze(B)
+
+np.diagonal(A)
+np.diagonal(B)
+
+np.trace(A)
+np.trace(B)
+
+np.ravel(a)
+np.ravel(b)
+np.ravel(c)
+np.ravel(A)
+np.ravel(B)
+
+np.nonzero(A)
+np.nonzero(B)
+
+np.shape(a)
+np.shape(b)
+np.shape(c)
+np.shape(A)
+np.shape(B)
+
+np.compress([True], a)
+np.compress([True], b)
+np.compress([True], c)
+np.compress([True], A)
+np.compress([True], B)
+
+np.clip(a, 0, 1.0)
+np.clip(b, -1, 1)
+np.clip(a, 0, None)
+np.clip(b, None, 1)
+np.clip(c, 0, 1)
+np.clip(A, 0, 1)
+np.clip(B, 0, 1)
+np.clip(B, [0, 1], [1, 2])
+
+np.sum(a)
+np.sum(b)
+np.sum(c)
+np.sum(A)
+np.sum(B)
+np.sum(A, axis=0)
+np.sum(B, axis=0)
+
+np.all(a)
+np.all(b)
+np.all(c)
+np.all(A)
+np.all(B)
+np.all(A, axis=0)
+np.all(B, axis=0)
+np.all(A, keepdims=True)
+np.all(B, keepdims=True)
+
+np.any(a)
+np.any(b)
+np.any(c)
+np.any(A)
+np.any(B)
+np.any(A, axis=0)
+np.any(B, axis=0)
+np.any(A, keepdims=True)
+np.any(B, keepdims=True)
+
+np.cumsum(a)
+np.cumsum(b)
+np.cumsum(c)
+np.cumsum(A)
+np.cumsum(B)
+
+np.ptp(b)
+np.ptp(c)
+np.ptp(B)
+np.ptp(B, axis=0)
+np.ptp(B, keepdims=True)
+
+np.amax(a)
+np.amax(b)
+np.amax(c)
+np.amax(A)
+np.amax(B)
+np.amax(A, axis=0)
+np.amax(B, axis=0)
+np.amax(A, keepdims=True)
+np.amax(B, keepdims=True)
+
+np.amin(a)
+np.amin(b)
+np.amin(c)
+np.amin(A)
+np.amin(B)
+np.amin(A, axis=0)
+np.amin(B, axis=0)
+np.amin(A, keepdims=True)
+np.amin(B, keepdims=True)
+
+np.prod(a)
+np.prod(b)
+np.prod(c)
+np.prod(A)
+np.prod(B)
+np.prod(a, dtype=None)
+np.prod(A, dtype=None)
+np.prod(A, axis=0)
+np.prod(B, axis=0)
+np.prod(A, keepdims=True)
+np.prod(B, keepdims=True)
+np.prod(b, out=d)
+np.prod(B, out=d)
+
+np.cumprod(a)
+np.cumprod(b)
+np.cumprod(c)
+np.cumprod(A)
+np.cumprod(B)
+
+np.ndim(a)
+np.ndim(b)
+np.ndim(c)
+np.ndim(A)
+np.ndim(B)
+
+np.size(a)
+np.size(b)
+np.size(c)
+np.size(A)
+np.size(B)
+
+np.around(a)
+np.around(b)
+np.around(c)
+np.around(A)
+np.around(B)
+
+np.mean(a)
+np.mean(b)
+np.mean(c)
+np.mean(A)
+np.mean(B)
+np.mean(A, axis=0)
+np.mean(B, axis=0)
+np.mean(A, keepdims=True)
+np.mean(B, keepdims=True)
+np.mean(b, out=d)
+np.mean(B, out=d)
+
+np.std(a)
+np.std(b)
+np.std(c)
+np.std(A)
+np.std(B)
+np.std(A, axis=0)
+np.std(B, axis=0)
+np.std(A, keepdims=True)
+np.std(B, keepdims=True)
+np.std(b, out=d)
+np.std(B, out=d)
+
+np.var(a)
+np.var(b)
+np.var(c)
+np.var(A)
+np.var(B)
+np.var(A, axis=0)
+np.var(B, axis=0)
+np.var(A, keepdims=True)
+np.var(B, keepdims=True)
+np.var(b, out=d)
+np.var(B, out=d)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/index_tricks.py

@@ -0,0 +1,64 @@
+from __future__ import annotations
+from typing import Any
+import numpy as np
+
+AR_LIKE_b = [[True, True], [True, True]]
+AR_LIKE_i = [[1, 2], [3, 4]]
+AR_LIKE_f = [[1.0, 2.0], [3.0, 4.0]]
+AR_LIKE_U = [["1", "2"], ["3", "4"]]
+
+AR_i8: np.ndarray[Any, np.dtype[np.int64]] = np.array(AR_LIKE_i, dtype=np.int64)
+
+np.ndenumerate(AR_i8)
+np.ndenumerate(AR_LIKE_f)
+np.ndenumerate(AR_LIKE_U)
+
+np.ndenumerate(AR_i8).iter
+np.ndenumerate(AR_LIKE_f).iter
+np.ndenumerate(AR_LIKE_U).iter
+
+next(np.ndenumerate(AR_i8))
+next(np.ndenumerate(AR_LIKE_f))
+next(np.ndenumerate(AR_LIKE_U))
+
+iter(np.ndenumerate(AR_i8))
+iter(np.ndenumerate(AR_LIKE_f))
+iter(np.ndenumerate(AR_LIKE_U))
+
+iter(np.ndindex(1, 2, 3))
+next(np.ndindex(1, 2, 3))
+
+np.unravel_index([22, 41, 37], (7, 6))
+np.unravel_index([31, 41, 13], (7, 6), order='F')
+np.unravel_index(1621, (6, 7, 8, 9))
+
+np.ravel_multi_index(AR_LIKE_i, (7, 6))
+np.ravel_multi_index(AR_LIKE_i, (7, 6), order='F')
+np.ravel_multi_index(AR_LIKE_i, (4, 6), mode='clip')
+np.ravel_multi_index(AR_LIKE_i, (4, 4), mode=('clip', 'wrap'))
+np.ravel_multi_index((3, 1, 4, 1), (6, 7, 8, 9))
+
+np.mgrid[1:1:2]
+np.mgrid[1:1:2, None:10]
+
+np.ogrid[1:1:2]
+np.ogrid[1:1:2, None:10]
+
+np.index_exp[0:1]
+np.index_exp[0:1, None:3]
+np.index_exp[0, 0:1, ..., [0, 1, 3]]
+
+np.s_[0:1]
+np.s_[0:1, None:3]
+np.s_[0, 0:1, ..., [0, 1, 3]]
+
+np.ix_(AR_LIKE_b[0])
+np.ix_(AR_LIKE_i[0], AR_LIKE_f[0])
+np.ix_(AR_i8[0])
+
+np.fill_diagonal(AR_i8, 5)
+
+np.diag_indices(4)
+np.diag_indices(2, 3)
+
+np.diag_indices_from(AR_i8)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/literal.py

@@ -0,0 +1,47 @@
+from __future__ import annotations
+
+from functools import partial
+from collections.abc import Callable
+
+import pytest  # type: ignore
+import numpy as np
+
+AR = np.array(0)
+AR.setflags(write=False)
+
+KACF = frozenset({None, "K", "A", "C", "F"})
+ACF = frozenset({None, "A", "C", "F"})
+CF = frozenset({None, "C", "F"})
+
+order_list: list[tuple[frozenset, Callable]] = [
+    (KACF, partial(np.ndarray, 1)),
+    (KACF, AR.tobytes),
+    (KACF, partial(AR.astype, int)),
+    (KACF, AR.copy),
+    (ACF, partial(AR.reshape, 1)),
+    (KACF, AR.flatten),
+    (KACF, AR.ravel),
+    (KACF, partial(np.array, 1)),
+    (CF, partial(np.zeros, 1)),
+    (CF, partial(np.ones, 1)),
+    (CF, partial(np.empty, 1)),
+    (CF, partial(np.full, 1, 1)),
+    (KACF, partial(np.zeros_like, AR)),
+    (KACF, partial(np.ones_like, AR)),
+    (KACF, partial(np.empty_like, AR)),
+    (KACF, partial(np.full_like, AR, 1)),
+    (KACF, partial(np.add, 1, 1)),  # i.e. np.ufunc.__call__
+    (ACF, partial(np.reshape, AR, 1)),
+    (KACF, partial(np.ravel, AR)),
+    (KACF, partial(np.asarray, 1)),
+    (KACF, partial(np.asanyarray, 1)),
+]
+
+for order_set, func in order_list:
+    for order in order_set:
+        func(order=order)
+
+    invalid_orders = KACF - order_set
+    for order in invalid_orders:
+        with pytest.raises(ValueError):
+            func(order=order)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/modules.py

@@ -0,0 +1,42 @@
+import numpy as np
+from numpy import f2py
+
+np.char
+np.ctypeslib
+np.emath
+np.fft
+np.lib
+np.linalg
+np.ma
+np.matrixlib
+np.polynomial
+np.random
+np.rec
+np.testing
+np.version
+
+np.lib.format
+np.lib.mixins
+np.lib.scimath
+np.lib.stride_tricks
+np.ma.extras
+np.polynomial.chebyshev
+np.polynomial.hermite
+np.polynomial.hermite_e
+np.polynomial.laguerre
+np.polynomial.legendre
+np.polynomial.polynomial
+
+np.__path__
+np.__version__
+
+np.__all__
+np.char.__all__
+np.ctypeslib.__all__
+np.emath.__all__
+np.lib.__all__
+np.ma.__all__
+np.random.__all__
+np.rec.__all__
+np.testing.__all__
+f2py.__all__

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/multiarray.py

@@ -0,0 +1,76 @@
+import numpy as np
+import numpy.typing as npt
+
+AR_f8: npt.NDArray[np.float64] = np.array([1.0])
+AR_i4 = np.array([1], dtype=np.int32)
+AR_u1 = np.array([1], dtype=np.uint8)
+
+AR_LIKE_f = [1.5]
+AR_LIKE_i = [1]
+
+b_f8 = np.broadcast(AR_f8)
+b_i4_f8_f8 = np.broadcast(AR_i4, AR_f8, AR_f8)
+
+next(b_f8)
+b_f8.reset()
+b_f8.index
+b_f8.iters
+b_f8.nd
+b_f8.ndim
+b_f8.numiter
+b_f8.shape
+b_f8.size
+
+next(b_i4_f8_f8)
+b_i4_f8_f8.reset()
+b_i4_f8_f8.ndim
+b_i4_f8_f8.index
+b_i4_f8_f8.iters
+b_i4_f8_f8.nd
+b_i4_f8_f8.numiter
+b_i4_f8_f8.shape
+b_i4_f8_f8.size
+
+np.inner(AR_f8, AR_i4)
+
+np.where([True, True, False])
+np.where([True, True, False], 1, 0)
+
+np.lexsort([0, 1, 2])
+
+np.can_cast(np.dtype("i8"), int)
+np.can_cast(AR_f8, "f8")
+np.can_cast(AR_f8, np.complex128, casting="unsafe")
+
+np.min_scalar_type([1])
+np.min_scalar_type(AR_f8)
+
+np.result_type(int, AR_i4)
+np.result_type(AR_f8, AR_u1)
+np.result_type(AR_f8, np.complex128)
+
+np.dot(AR_LIKE_f, AR_i4)
+np.dot(AR_u1, 1)
+np.dot(1.5j, 1)
+np.dot(AR_u1, 1, out=AR_f8)
+
+np.vdot(AR_LIKE_f, AR_i4)
+np.vdot(AR_u1, 1)
+np.vdot(1.5j, 1)
+
+np.bincount(AR_i4)
+
+np.copyto(AR_f8, [1.6])
+
+np.putmask(AR_f8, [True], 1.5)
+
+np.packbits(AR_i4)
+np.packbits(AR_u1)
+
+np.unpackbits(AR_u1)
+
+np.shares_memory(1, 2)
+np.shares_memory(AR_f8, AR_f8, max_work=1)
+
+np.may_share_memory(1, 2)
+np.may_share_memory(AR_f8, AR_f8, max_work=1)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/ndarray_conversion.py

@@ -0,0 +1,94 @@
+import os
+import tempfile
+
+import numpy as np
+
+nd = np.array([[1, 2], [3, 4]])
+scalar_array = np.array(1)
+
+# item
+scalar_array.item()
+nd.item(1)
+nd.item(0, 1)
+nd.item((0, 1))
+
+# tolist is pretty simple
+
+# itemset
+scalar_array.itemset(3)
+nd.itemset(3, 0)
+nd.itemset((0, 0), 3)
+
+# tobytes
+nd.tobytes()
+nd.tobytes("C")
+nd.tobytes(None)
+
+# tofile
+if os.name != "nt":
+    with tempfile.NamedTemporaryFile(suffix=".txt") as tmp:
+        nd.tofile(tmp.name)
+        nd.tofile(tmp.name, "")
+        nd.tofile(tmp.name, sep="")
+
+        nd.tofile(tmp.name, "", "%s")
+        nd.tofile(tmp.name, format="%s")
+
+        nd.tofile(tmp)
+
+# dump is pretty simple
+# dumps is pretty simple
+
+# astype
+nd.astype("float")
+nd.astype(float)
+
+nd.astype(float, "K")
+nd.astype(float, order="K")
+
+nd.astype(float, "K", "unsafe")
+nd.astype(float, casting="unsafe")
+
+nd.astype(float, "K", "unsafe", True)
+nd.astype(float, subok=True)
+
+nd.astype(float, "K", "unsafe", True, True)
+nd.astype(float, copy=True)
+
+# byteswap
+nd.byteswap()
+nd.byteswap(True)
+
+# copy
+nd.copy()
+nd.copy("C")
+
+# view
+nd.view()
+nd.view(np.int64)
+nd.view(dtype=np.int64)
+nd.view(np.int64, np.matrix)
+nd.view(type=np.matrix)
+
+# getfield
+complex_array = np.array([[1 + 1j, 0], [0, 1 - 1j]], dtype=np.complex128)
+
+complex_array.getfield("float")
+complex_array.getfield(float)
+
+complex_array.getfield("float", 8)
+complex_array.getfield(float, offset=8)
+
+# setflags
+nd.setflags()
+
+nd.setflags(True)
+nd.setflags(write=True)
+
+nd.setflags(True, True)
+nd.setflags(write=True, align=True)
+
+nd.setflags(True, True, False)
+nd.setflags(write=True, align=True, uic=False)
+
+# fill is pretty simple

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/scalars.py

@@ -0,0 +1,248 @@
+import sys
+import datetime as dt
+
+import pytest
+import numpy as np
+
+b =  np.bool_()
+u8 = np.uint64()
+i8 = np.int64()
+f8 = np.float64()
+c16 = np.complex128()
+U = np.str_()
+S = np.bytes_()
+
+
+# Construction
+class D:
+    def __index__(self) -> int:
+        return 0
+
+
+class C:
+    def __complex__(self) -> complex:
+        return 3j
+
+
+class B:
+    def __int__(self) -> int:
+        return 4
+
+
+class A:
+    def __float__(self) -> float:
+        return 4.0
+
+
+np.complex64(3j)
+np.complex64(A())
+np.complex64(C())
+np.complex128(3j)
+np.complex128(C())
+np.complex128(None)
+np.complex64("1.2")
+np.complex128(b"2j")
+
+np.int8(4)
+np.int16(3.4)
+np.int32(4)
+np.int64(-1)
+np.uint8(B())
+np.uint32()
+np.int32("1")
+np.int64(b"2")
+
+np.float16(A())
+np.float32(16)
+np.float64(3.0)
+np.float64(None)
+np.float32("1")
+np.float16(b"2.5")
+
+np.uint64(D())
+np.float32(D())
+np.complex64(D())
+
+np.bytes_(b"hello")
+np.bytes_("hello", 'utf-8')
+np.bytes_("hello", encoding='utf-8')
+np.str_("hello")
+np.str_(b"hello", 'utf-8')
+np.str_(b"hello", encoding='utf-8')
+
+# Array-ish semantics
+np.int8().real
+np.int16().imag
+np.int32().data
+np.int64().flags
+
+np.uint8().itemsize * 2
+np.uint16().ndim + 1
+np.uint32().strides
+np.uint64().shape
+
+# Time structures
+np.datetime64()
+np.datetime64(0, "D")
+np.datetime64(0, b"D")
+np.datetime64(0, ('ms', 3))
+np.datetime64("2019")
+np.datetime64(b"2019")
+np.datetime64("2019", "D")
+np.datetime64(np.datetime64())
+np.datetime64(dt.datetime(2000, 5, 3))
+np.datetime64(dt.date(2000, 5, 3))
+np.datetime64(None)
+np.datetime64(None, "D")
+
+np.timedelta64()
+np.timedelta64(0)
+np.timedelta64(0, "D")
+np.timedelta64(0, ('ms', 3))
+np.timedelta64(0, b"D")
+np.timedelta64("3")
+np.timedelta64(b"5")
+np.timedelta64(np.timedelta64(2))
+np.timedelta64(dt.timedelta(2))
+np.timedelta64(None)
+np.timedelta64(None, "D")
+
+np.void(1)
+np.void(np.int64(1))
+np.void(True)
+np.void(np.bool_(True))
+np.void(b"test")
+np.void(np.bytes_("test"))
+np.void(object(), [("a", "O"), ("b", "O")])
+np.void(object(), dtype=[("a", "O"), ("b", "O")])
+
+# Protocols
+i8 = np.int64()
+u8 = np.uint64()
+f8 = np.float64()
+c16 = np.complex128()
+b_ = np.bool_()
+td = np.timedelta64()
+U = np.str_("1")
+S = np.bytes_("1")
+AR = np.array(1, dtype=np.float64)
+
+int(i8)
+int(u8)
+int(f8)
+int(b_)
+int(td)
+int(U)
+int(S)
+int(AR)
+with pytest.warns(np.ComplexWarning):
+    int(c16)
+
+float(i8)
+float(u8)
+float(f8)
+float(b_)
+float(td)
+float(U)
+float(S)
+float(AR)
+with pytest.warns(np.ComplexWarning):
+    float(c16)
+
+complex(i8)
+complex(u8)
+complex(f8)
+complex(c16)
+complex(b_)
+complex(td)
+complex(U)
+complex(AR)
+
+
+# Misc
+c16.dtype
+c16.real
+c16.imag
+c16.real.real
+c16.real.imag
+c16.ndim
+c16.size
+c16.itemsize
+c16.shape
+c16.strides
+c16.squeeze()
+c16.byteswap()
+c16.transpose()
+
+# Aliases
+np.string_()
+
+np.byte()
+np.short()
+np.intc()
+np.intp()
+np.int_()
+np.longlong()
+
+np.ubyte()
+np.ushort()
+np.uintc()
+np.uintp()
+np.uint()
+np.ulonglong()
+
+np.half()
+np.single()
+np.double()
+np.float_()
+np.longdouble()
+np.longfloat()
+
+np.csingle()
+np.singlecomplex()
+np.cdouble()
+np.complex_()
+np.cfloat()
+np.clongdouble()
+np.clongfloat()
+np.longcomplex()
+
+b.item()
+i8.item()
+u8.item()
+f8.item()
+c16.item()
+U.item()
+S.item()
+
+b.tolist()
+i8.tolist()
+u8.tolist()
+f8.tolist()
+c16.tolist()
+U.tolist()
+S.tolist()
+
+b.ravel()
+i8.ravel()
+u8.ravel()
+f8.ravel()
+c16.ravel()
+U.ravel()
+S.ravel()
+
+b.flatten()
+i8.flatten()
+u8.flatten()
+f8.flatten()
+c16.flatten()
+U.flatten()
+S.flatten()
+
+b.reshape(1)
+i8.reshape(1)
+u8.reshape(1)
+f8.reshape(1)
+c16.reshape(1)
+U.reshape(1)
+S.reshape(1)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/ufunclike.py

@@ -0,0 +1,46 @@
+from __future__ import annotations
+from typing import Any
+import numpy as np
+
+
+class Object:
+    def __ceil__(self) -> Object:
+        return self
+
+    def __floor__(self) -> Object:
+        return self
+
+    def __ge__(self, value: object) -> bool:
+        return True
+
+    def __array__(self) -> np.ndarray[Any, np.dtype[np.object_]]:
+        ret = np.empty((), dtype=object)
+        ret[()] = self
+        return ret
+
+
+AR_LIKE_b = [True, True, False]
+AR_LIKE_u = [np.uint32(1), np.uint32(2), np.uint32(3)]
+AR_LIKE_i = [1, 2, 3]
+AR_LIKE_f = [1.0, 2.0, 3.0]
+AR_LIKE_O = [Object(), Object(), Object()]
+AR_U: np.ndarray[Any, np.dtype[np.str_]] = np.zeros(3, dtype="U5")
+
+np.fix(AR_LIKE_b)
+np.fix(AR_LIKE_u)
+np.fix(AR_LIKE_i)
+np.fix(AR_LIKE_f)
+np.fix(AR_LIKE_O)
+np.fix(AR_LIKE_f, out=AR_U)
+
+np.isposinf(AR_LIKE_b)
+np.isposinf(AR_LIKE_u)
+np.isposinf(AR_LIKE_i)
+np.isposinf(AR_LIKE_f)
+np.isposinf(AR_LIKE_f, out=AR_U)
+
+np.isneginf(AR_LIKE_b)
+np.isneginf(AR_LIKE_u)
+np.isneginf(AR_LIKE_i)
+np.isneginf(AR_LIKE_f)
+np.isneginf(AR_LIKE_f, out=AR_U)

LTA_openwebtext_dualt/mini_owt_logdirichlet/.venv_qwen35_uv/lib/python3.12/site-packages/numpy/typing/tests/data/pass/warnings_and_errors.py

@@ -0,0 +1,6 @@
+import numpy as np
+
+np.AxisError("test")
+np.AxisError(1, ndim=2)
+np.AxisError(1, ndim=2, msg_prefix="error")
+np.AxisError(1, ndim=2, msg_prefix=None)