Publish 1903 files, replacing 1315 cells

README.md

@@ -228,7 +228,7 @@ Start from `tags.parquet`: one row per (`tag`, `predictor`, file), with the
 model, method, configuration, storage class, layer and provenance of every
 cell decoded, so a cell can be found by model, method and configuration.
 
-- 2,596 cells across 58 corpora, 280,030,555 scalar rows
+- 2,596 cells across 58 corpora, 280,034,199 scalar rows
 - 25 methods, 26 models and frequency tables
 
 ## Layout
@@ -283,7 +283,7 @@ A `positional` estimator conditions on a prefix ending at a token position and r
 | `unit_rule` | cells |
 |---|---|
 | `spacy_sentence` | 26 |
-| `whitespace_continuation` | 395 |
+| `whitespace_continuation` | 403 |
 
 ## Word values derived from token tables
 
@@ -352,14 +352,14 @@ Each recompute wave was checked table by table against the corpus loader's unit
 |---|---|---|
 | `verification/wave1-tokens.json` | check | 406 |
 | `verification/wave1-tokens.json` | pass | 6 |
-| `verification/wave2.json` | check | 269 |
-| `verification/wave2.json` | pass | 626 |
+| `verification/wave2.json` | check | 272 |
+| `verification/wave2.json` | pass | 631 |
 
 | flag | tables |
 |---|---|
 | manifest identity lacks storage_class | 405 |
 | manifest layer is None | 405 |
-| words empty or containing whitespace | 274 |
+| words empty or containing whitespace | 277 |
 | context scope story (layer 1 conditions on the whole stimulus) | 49 |
 | BOS disabled | 42 |
 | context scope story | 28 |
@@ -380,13 +380,13 @@ Each recompute wave was checked table by table against the corpus loader's unit
 
 ## Coverage and gaps
 
-`coverage.parquet` has one row per planned (method, model, corpus, unit) table and one per produced table no plan names. `status` is `exported` when the table comes from the latest release, `carried` when the store ships it from an earlier one, `quarantined` when the compute driver refused to publish it (a stimulus failed or the values failed an audit) and the store has no earlier table, and `absent` when the store has none. `reason_class` is the plan's class for the table, one of `complete`, `excluded_by_decision`, `expected_refusal`, `held_decision`, `known_gap_beam_empty`, `missing`, `running`. `expected_refusal` is a whitespace-word estimator on a corpus whose words are not the model's words, and `known_gap_beam_empty` is a character beam that emptied on a stimulus. `detail` holds the failure log of a quarantined table or the command of an absent one.
+`coverage.parquet` has one row per planned (method, model, corpus, unit) table and one per produced table no plan names. `status` is `exported` when the table comes from the latest release, `carried` when the store ships it from an earlier one, `quarantined` when the compute driver refused to publish it (a stimulus failed or the values failed an audit) and the store has no earlier table, and `absent` when the store has none. `reason_class` is the plan's class for the table, one of `complete`, `excluded_by_decision`, `expected_refusal`, `held_decision`, `known_gap_beam_empty`, `missing`. `expected_refusal` is a whitespace-word estimator on a corpus whose words are not the model's words, and `known_gap_beam_empty` is a character beam that emptied on a stimulus. `detail` holds the failure log of a quarantined table or the command of an absent one.
 
 | `status` | tables |
 |---|---|
 | absent | 1 |
-| carried | 43 |
-| exported | 1309 |
+| carried | 35 |
+| exported | 1317 |
 | quarantined | 5 |
 
 | `status` | `reason_class` | tables |
@@ -396,9 +396,8 @@ Each recompute wave was checked table by table against the corpus loader's unit
 | carried | expected_refusal | 6 |
 | carried | held_decision | 24 |
 | carried | known_gap_beam_empty | 2 |
-| carried | missing | 8 |
-| carried | running | 2 |
-| exported | complete | 855 |
+| carried | missing | 2 |
+| exported | complete | 863 |
 | exported | expected_refusal | 2 |
 | exported | not planned | 452 |
 | quarantined | held_decision | 5 |
@@ -411,39 +410,39 @@ Each recompute wave was checked table by table against the corpus loader's unit
 
 ## What changed in this release
 
-Built on the published store at commit `2508a73d9431d94a3b776ced13c3fc44f49d9427`; every published cell this release does not replace is carried over unchanged.
+Built on the published store at commit `8084551878470d9b11de74c12f172ed3c0717206`; every published cell this release does not replace is carried over unchanged.
 
 | wave | cells exported |
 |---|---|
 | wave1-tokens | 412 |
-| wave2 | 895 |
+| wave2 | 903 |
 
-- 825 published cells replaced by a recomputed table of the same tag
+- 1315 published cells replaced by a recomputed table of the same tag
 - 0 published files deleted because their cell was recomputed without the array they held
-- 1895 files written
+- 1903 files written
 - new: `tokens/` (token-grain tables), `characters/` (character-grain tables), `coverage.parquet`, `verification/`, and the storage columns of the word tables and the index
 - array keys without a store folder, left in the pickles only: `orth_similarity_samples`, `pos_match_samples`, `sampled_words`, `similarity_samples_tokens`
 
 ### Carried cells on an older unit grid
 
-133 cells this release carries were computed against an earlier version of their corpus, so their `stimulus_id` and `unit_index` do not line up with the cells this release stages into the same table. Join them to a recomputed predictor only after checking the units match. `tags.parquet` gives each cell's `n_stimuli` and `n_rows`, and a recomputed cell of the same corpus gives the current shape.
+128 cells this release carries were computed against an earlier version of their corpus, so their `stimulus_id` and `unit_index` do not line up with the cells this release stages into the same table. Join them to a recomputed predictor only after checking the units match. `tags.parquet` gives each cell's `n_stimuli` and `n_rows`, and a recomputed cell of the same corpus gives the current shape.
 
 | table | cells on an older grid |
 |---|---|
-| `dundee.parquet` | 20 |
+| `dundee.parquet` | 19 |
 | `meco_de.parquet` | 2 |
-| `meco_ee.parquet` | 7 |
+| `meco_ee.parquet` | 6 |
 | `meco_fi.parquet` | 5 |
-| `meco_gr.parquet` | 6 |
+| `meco_gr.parquet` | 5 |
 | `meco_it.parquet` | 5 |
 | `meco_l1_w2_ch_s.parquet` | 6 |
 | `meco_l1_w2_ch_t.parquet` | 6 |
 | `meco_l2_w2.parquet` | 10 |
 | `meco_tr.parquet` | 5 |
-| `onestop.parquet` | 17 |
+| `onestop.parquet` | 16 |
 | `provo.parquet` | 15 |
 | `sbsat.parquet` | 20 |
-| `szewczyk_n400.parquet` | 9 |
+| `szewczyk_n400.parquet` | 8 |
 
 ## Loading
 

celer.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bdd50cec403dddae96d386f1ffbf4e1131d1235a0308caad6125098656adfb09
-size 175160031
+oid sha256:c80e5e466e6e56d3faa7514ab1aab401f15cfdeb2064bb66bbaf21691064856b
+size 174977391

coverage.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9be70c85eba239bbd47947709627a6ee84e70859790bb5b33bcc21ef900cddfa
-size 16776
+oid sha256:e1ea2d6ee9dfaab9766ee72dd4c1ec45cb57d40c29a4ab3faa873d4ca8d1ba20
+size 16738

dundee.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3b955bdc56e59053f025ec3a3bb3c81d3acbd608bd68f080d787c2bc073d280
-size 41711159
+oid sha256:a522e1f64f8cddd90b4c07e0b10dce6d0ab98cab79d48ddb370f5190dcc6151c
+size 40819012

federmeier2007_n400.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a952fe2079d20cc7bb8f30e7be5446ec232f7d331a53ab13183692539653a310
-size 2539483
+oid sha256:97e6520742bb30495d9000a675770033b3e608dc794d76b8dee39b9c6e362793
+size 2532477

geco.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e2e81992ee05425d51962b6141885b8cda1479416da5de8b186481e34867ab5
-size 47410281
+oid sha256:34742f073defd3cbf4af4e2ddbaa52e3a2f6c937af3de51a5f2833d700a3550d
+size 46930884

history.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2d5361166e5bfbce7fc8cc7061fc47e29cc6789dcf25325a6b20e9698cd9d7f5
-size 29293
+oid sha256:e6f366132adfa50ae2a3939d95fd25d5aadb6ce46e83ebfb5aa3371ebf650701
+size 29479

manifests/ai-forever_mGPT_mmc_unigram_surprisal_meco_ee.json

@@ -0,0 +1,69 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "ai-forever/mGPT",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "et",
+      "language_override": null,
+      "chunk_size": 64,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "ai-forever/mGPT",
+      "revision": null,
+      "sha256": "36faadfd737c580e4d1d02f21e6b9ae4a429237a23d34b558694c9a059494c04",
+      "files": [
+        "config.json",
+        "merges.txt",
+        "pytorch_model.bin",
+        "special_tokens_map.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ],
+      "requested_revision": null
+    },
+    "corpus_sha256": "0e862f2c2aee076712ef53ecdfd5b8e32e38fa2122af234f6438499bcffd9177",
+    "n_stimuli": 12,
+    "source_sha256": "6970c751bdbd4585eea9809f91a75caa42891b639d2ce8aabb2f5e8f1844aa5b",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "021b035370d1a7cc1a8e7bd34b7e505c58ce3d6e30b1c8bd648657c80d1247cd",
+  "environment": {
+    "code_commit": "052babe864ba0b391c811d2d28cc2fbe04ac42f0",
+    "code_dirty": false,
+    "device": "NVIDIA L40S",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-25T19:31:49+00:00"
+  }
+}

manifests/ai-forever_mGPT_mmc_unigram_surprisal_meco_gr.json

@@ -0,0 +1,69 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "ai-forever/mGPT",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "el",
+      "language_override": null,
+      "chunk_size": 64,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "ai-forever/mGPT",
+      "revision": null,
+      "sha256": "36faadfd737c580e4d1d02f21e6b9ae4a429237a23d34b558694c9a059494c04",
+      "files": [
+        "config.json",
+        "merges.txt",
+        "pytorch_model.bin",
+        "special_tokens_map.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ],
+      "requested_revision": null
+    },
+    "corpus_sha256": "bb46487d3d7a77783fe5fe5c1499a0b8098eead45b44fae5e3c113196e174992",
+    "n_stimuli": 12,
+    "source_sha256": "6970c751bdbd4585eea9809f91a75caa42891b639d2ce8aabb2f5e8f1844aa5b",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "1d6ad9b958a51a511b0b4a91eb236c0927484e9700d1d0361b93a3108e59fcd0",
+  "environment": {
+    "code_commit": "052babe864ba0b391c811d2d28cc2fbe04ac42f0",
+    "code_dirty": false,
+    "device": "NVIDIA L40S",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-25T12:17:23+00:00"
+  }
+}

manifests/ai-forever_mGPT_mmc_unigram_surprisal_meco_ko.json

@@ -0,0 +1,69 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "ai-forever/mGPT",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "ko",
+      "language_override": null,
+      "chunk_size": null,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "ai-forever/mGPT",
+      "revision": null,
+      "sha256": "36faadfd737c580e4d1d02f21e6b9ae4a429237a23d34b558694c9a059494c04",
+      "files": [
+        "config.json",
+        "merges.txt",
+        "pytorch_model.bin",
+        "special_tokens_map.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ],
+      "requested_revision": null
+    },
+    "corpus_sha256": "75b94d84721674538046087d46de51f3faa9a774c6cd8681dd5c04fd1ef0cde5",
+    "n_stimuli": 12,
+    "source_sha256": "6970c751bdbd4585eea9809f91a75caa42891b639d2ce8aabb2f5e8f1844aa5b",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "8509b3c18eda39f7e435516cec75495a59d5d5ebf225ea6af498099d6aa9af3f",
+  "environment": {
+    "code_commit": "052babe864ba0b391c811d2d28cc2fbe04ac42f0",
+    "code_dirty": false,
+    "device": "NVIDIA L40S",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-25T12:48:00+00:00"
+  }
+}

manifests/ai-forever_mGPT_mmc_unigram_surprisal_szewczyk_n400.json

@@ -0,0 +1,69 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "ai-forever/mGPT",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "en",
+      "language_override": null,
+      "chunk_size": 64,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "ai-forever/mGPT",
+      "revision": null,
+      "sha256": "36faadfd737c580e4d1d02f21e6b9ae4a429237a23d34b558694c9a059494c04",
+      "files": [
+        "config.json",
+        "merges.txt",
+        "pytorch_model.bin",
+        "special_tokens_map.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ],
+      "requested_revision": null
+    },
+    "corpus_sha256": "3310e0ab23d763a8845582e6037a0f7cc8c045bf03766db292ca0fa5af04b078",
+    "n_stimuli": 1331,
+    "source_sha256": "40fd4bc86f96809413e4945f69d1d59272b2de15ac949a4e90f80ec4e90434d4",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "0d550cbea503a231ff990c8c500f26adc6a9291b5fc99a0d0bcd8fa46b6d371f",
+  "environment": {
+    "code_commit": "e78a32c204cc3589c964824ba60b43c1ff2bde8b",
+    "code_dirty": false,
+    "device": "NVIDIA L40S",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-24T05:45:13+00:00"
+  }
+}

manifests/gpt2_mmc_unigram_surprisal_dundee.json

@@ -0,0 +1,77 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "gpt2",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "en",
+      "language_override": null,
+      "chunk_size": null,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "gpt2",
+      "revision": null,
+      "sha256": "c5a644aeb1cf4cde835b077c20075efa570b2b51cec32626f8008b4fe580107e",
+      "files": [
+        "config.json",
+        "generation_config.json",
+        "merges.txt",
+        "model.safetensors",
+        "onnx/config.json",
+        "onnx/generation_config.json",
+        "onnx/merges.txt",
+        "onnx/special_tokens_map.json",
+        "onnx/tokenizer.json",
+        "onnx/tokenizer_config.json",
+        "onnx/vocab.json",
+        "tokenizer.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ],
+      "requested_revision": null
+    },
+    "corpus_sha256": "7e634bbd21030e4d5386193726ebf74957ae28f4555911a655154211cd9f8d88",
+    "n_stimuli": 20,
+    "source_sha256": "40fd4bc86f96809413e4945f69d1d59272b2de15ac949a4e90f80ec4e90434d4",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "5a5eca17a09118e6a7146f0a377bc9d582d6eb32d9c91d1e0ba1dfbd3282a8ed",
+  "environment": {
+    "code_commit": "e78a32c204cc3589c964824ba60b43c1ff2bde8b",
+    "code_dirty": false,
+    "device": "NVIDIA L40S",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-23T23:31:34+00:00"
+  }
+}

manifests/gpt2_mmc_unigram_surprisal_geco.json

@@ -0,0 +1,76 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "gpt2",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "en",
+      "language_override": null,
+      "chunk_size": 64,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "openai-community/gpt2",
+      "revision": "607a30d783dfa663caf39e06633721c8d4cfcd7e",
+      "sha256": "c5a644aeb1cf4cde835b077c20075efa570b2b51cec32626f8008b4fe580107e",
+      "files": [
+        "config.json",
+        "generation_config.json",
+        "merges.txt",
+        "model.safetensors",
+        "onnx/config.json",
+        "onnx/generation_config.json",
+        "onnx/merges.txt",
+        "onnx/special_tokens_map.json",
+        "onnx/tokenizer.json",
+        "onnx/tokenizer_config.json",
+        "onnx/vocab.json",
+        "tokenizer.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ]
+    },
+    "corpus_sha256": "29d759acef49c2eb2ab44d396f83853833e24ba5fcbaef7c50132819fc6aab95",
+    "n_stimuli": 588,
+    "source_sha256": "40fd4bc86f96809413e4945f69d1d59272b2de15ac949a4e90f80ec4e90434d4",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "5878d02db613654f1165d403e48b012b244d76ad591312c00228ed6e397be21f",
+  "environment": {
+    "code_commit": "e78a32c204cc3589c964824ba60b43c1ff2bde8b",
+    "code_dirty": false,
+    "device": "NVIDIA GeForce RTX 2080 Ti",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-26T10:16:22+00:00"
+  }
+}

manifests/gpt2_mmc_unigram_surprisal_onestop.json

@@ -0,0 +1,76 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "mmc_unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "gpt2",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "en",
+      "language_override": null,
+      "chunk_size": 64,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "openai-community/gpt2",
+      "revision": "607a30d783dfa663caf39e06633721c8d4cfcd7e",
+      "sha256": "c5a644aeb1cf4cde835b077c20075efa570b2b51cec32626f8008b4fe580107e",
+      "files": [
+        "config.json",
+        "generation_config.json",
+        "merges.txt",
+        "model.safetensors",
+        "onnx/config.json",
+        "onnx/generation_config.json",
+        "onnx/merges.txt",
+        "onnx/special_tokens_map.json",
+        "onnx/tokenizer.json",
+        "onnx/tokenizer_config.json",
+        "onnx/vocab.json",
+        "tokenizer.json",
+        "tokenizer_config.json",
+        "vocab.json"
+      ]
+    },
+    "corpus_sha256": "304cd0369bc96952403c2fbf6a01bad2a8f045b24f82bbcf6309f1eab78aab8d",
+    "n_stimuli": 339,
+    "source_sha256": "40fd4bc86f96809413e4945f69d1d59272b2de15ac949a4e90f80ec4e90434d4",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "76156a1b63bfd4bbbafa2603d44189a07a333b542eea901d7eb2cdb7793e7add",
+  "environment": {
+    "code_commit": "e78a32c204cc3589c964824ba60b43c1ff2bde8b",
+    "code_dirty": false,
+    "device": "NVIDIA GeForce RTX 2080 Ti",
+    "model_device": null,
+    "cuda": "13.0",
+    "computed_at": "2026-09-25T02:56:48+00:00"
+  }
+}

manifests/wordfreq_unigram_surprisal_celer.json

@@ -0,0 +1,59 @@
+{
+  "identity": {
+    "schema_version": 1,
+    "method": "unigram_surprisal",
+    "storage_class": "unit_bound",
+    "layer": 1,
+    "options": {
+      "model_id": "wordfreq",
+      "beam_width": 5,
+      "n_samples": 64,
+      "max_tokens": 10,
+      "alpha": 1.0,
+      "seed": null,
+      "add_bos": true,
+      "dataset_language": "en",
+      "language_override": null,
+      "chunk_size": null,
+      "word_starters": [],
+      "top_p": 0.9,
+      "context_scope": "stimulus",
+      "unit": "words",
+      "window_policy": "reference",
+      "segmentation": "whitespace",
+      "model_revision": null,
+      "store_distribution": false,
+      "model_identity": null
+    },
+    "model": {
+      "source": "wordfreq",
+      "revision": "package-version"
+    },
+    "corpus_sha256": "7ec1058e5169feed00e160622998671316bba0f47990e35c048f55e66e65fabf",
+    "n_stimuli": 28535,
+    "source_sha256": "291aac7b6499cf9c4848744cff8542ce871144cfc25a1e18ddae922ca5d13850",
+    "versions": {
+      "python": "3.12.13",
+      "replica": "0.0.1",
+      "numpy": "2.4.4",
+      "torch": "2.11.0",
+      "transformers": "4.57.6",
+      "tokenizers": "0.22.2",
+      "wordfreq": "3.1.1",
+      "wordsprobability": "0.17",
+      "genlm-bytes": "0.2.0"
+    },
+    "unit_rule": "whitespace_continuation",
+    "unit_layer": "words"
+  },
+  "status": "completed",
+  "payload_sha256": "f5b88868712b9e82e3b13649256afdeef8c757755f73b99e8e2d5b1a5b613531",
+  "environment": {
+    "code_commit": "f8fc311dce61f277503c49e58f73902410d62708",
+    "code_dirty": false,
+    "device": "cpu",
+    "model_device": null,
+    "cuda": null,
+    "computed_at": "2026-09-23T17:40:24+00:00"
+  }
+}

meco_ee.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:05981999893c59fc6462ce8518af83458426f07da552eb30270e7aa8211eb423
-size 935950
+oid sha256:989f1c9bb336bd62890bceeaef14fcab8415727444ca5eb0f13fe8d5c21ae415
+size 943561

meco_gr.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bc1fd5a9a5284de9f60a197695ac44771aaa724ad252674e829a527240840e99
-size 1287723
+oid sha256:9348a1e1e6d0b5bf8180a8b7c7cefd5abb1cf09dafc13b514f55296ed02172aa
+size 1296454

meco_ko.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f4bc0f7cd310e8d87bebffc56033c539a8210f5955ac2e1619b324fcbf6a34f4
-size 1166021
+oid sha256:7dfbf4b24f5ff9ee08e7a29bd7175dda221142adc95dd5b730181e082c0435fc
+size 1172521

meco_l2_w2.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:97b21c6caa1dfa9a7018ee34b192b29d900c3f17820c4812cfbc5321136ac815
-size 1386632
+oid sha256:aaed7c5d695235011074c799e362b3ba16e4f2f3f12e1490e59c22f4e1c0c2e3
+size 1380590

natural_stories.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7181cd761a15868a5154e2e2a2ddf1569dcb01d48d0d8536ed11d4b84542151a
-size 9064134
+oid sha256:f306790204b7e151c26945bc65d5abb34a300bcdb511cea907f363e95e5ba23e
+size 9082557

natural_stories_maze.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d56abb2b0a45ae13f11f7d07dfc6cf2c21662d81b4930e677ac11f9250f5cb7
-size 4990536
+oid sha256:c9b8cbcc508b6ce64973236bd6d9d7847faae3ba00f72de29c787b5cddd48934
+size 4983501

onestop.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:319adbba366627305a4d5a83bf1b8d6171238a3caa59774adae9bdbaf03dd686
-size 25977821
+oid sha256:a4f5595a0f87ebfa860f083a32735212b7cf3657f8ee5f51d3b121b4832e8f5f
+size 25402682

provo.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21d5436a6a9c55b6e7196941aceaa591e85356660dfe0b4ace26eef1b02c7f61
-size 2742604
+oid sha256:bc2059c409e9f11bd68ae88158e459988aef91d462323db3eadf13c854127bfd
+size 2735079

sbsat.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:76693d3e1d7959200edf92e3c9a714591078f29f9318d1bc1812597aa7ca4895
-size 3491927
+oid sha256:03a20429aa1bc8f1acbd84ba854a47debcd53d0749ad66849ff15d4344b969b7
+size 3489192

szewczyk_federmeier2022_n400.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:98d95e863e5c6b3d064cffbd7de5c2687f7a3cc6eedac4b1f3c90a8229791574
-size 2428112
+oid sha256:41edacf0354b9e0d7388185094dbb4145e1ef651f464d5e1a18c45cde1602c89
+size 2420333

szewczyk_n400.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d8d84b2a8c446ad98b9afbc621d62e620e616d2177f7728625ca9a9237d0eb1
-size 4512201
+oid sha256:78ab00d99fb29c5a73923b3ee82030971b4cf9beaf737c3b52e5bb3cd86fbc69
+size 4501730

tags.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41fcd4e6438f3a410ac2a131fc010e6023a9821ff7a27934f584200ba57d5b6a
-size 73464
+oid sha256:c77b7e229d577b740ccf26d63136dcadc226acd8a797ac90fb7839355aef148f
+size 73658

verification/wave2.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5afb4037c9054933030357ee04afab2aaa5de02d290249eadf158a1a9f938146
-size 38674652
+oid sha256:9457dd47dd834ac8c3388424e792abbae70f751f5594c9dd072af231cf6b7767
+size 38982658

wlotko_federmeier2012_n400.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e14003e7d30937ba5b601aef83b9cbf037fe6f2e32295b7de00a00dbba525878
-size 1921908
+oid sha256:961f92b1309b8ea2126c1ef818786a0f94e2357116617acee4410e4d80345e8b
+size 1914917

zuco1.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c5f172501dbbf4f1f4bcb575cd2d0c8035019b6fa70bc1af4e1e184b98a6d61
-size 3556452
+oid sha256:a9b09b0371a98aa00b88b4e6f47bebb88a6d75f7150efa7d8cd8c531a931f07c
+size 3543148