lhallee commited on
Commit
c991ce5
·
verified ·
1 Parent(s): 2747554

Update FastPLMs files

Browse files
README.md CHANGED
@@ -124,11 +124,11 @@ Set `output` and `format="safetensors"` or `"sqlite"` for transactional,
124
  bounded-memory storage. Resume checks input order, model state, tokenizer
125
  policy, backend, dtype, and pooling configuration before it appends data.
126
 
127
- ## Downstream classification
128
 
129
- Both downstream AutoClasses use the checkpoint backbone and create a new,
130
- untrained `classifier`. Sequence labels have shape `(b,)`. Residue labels have
131
- shape `(b, l)` and use `-100` outside biological positions:
132
 
133
  ```python
134
  import torch
@@ -275,8 +275,8 @@ publication requires all artifact, legal, parity, and atomic-publication checks.
275
  ## Release record
276
 
277
  - FastPLMs weights: `Synthyra/DPLM-3B`
278
- - Runtime revision: recorded in the built artifact and published commit
279
- - Source-tree and runtime-bundle SHA-256: recorded in the source record
280
  - Official checkpoint: `airkingbd/dplm_3b`
281
  - Artifact source: `fast`
282
  - State transform: `dplm_to_fastplms_v1`
 
124
  bounded-memory storage. Resume checks input order, model state, tokenizer
125
  policy, backend, dtype, and pooling configuration before it appends data.
126
 
127
+ ## Downstream prediction
128
 
129
+ The sequence and token prediction AutoClasses use the checkpoint backbone and
130
+ create a new, untrained `classifier`. Sequence labels have shape `(b,)`.
131
+ Residue labels have shape `(b, l)` and use `-100` outside biological positions.
132
 
133
  ```python
134
  import torch
 
275
  ## Release record
276
 
277
  - FastPLMs weights: `Synthyra/DPLM-3B`
278
+ - Runtime revision: recorded separately in the built artifact and published commit
279
+ - Runtime source identities: recorded in `source-record.json`
280
  - Official checkpoint: `airkingbd/dplm_3b`
281
  - Artifact source: `fast`
282
  - State transform: `dplm_to_fastplms_v1`
fastplms/models.toml CHANGED
@@ -199,7 +199,7 @@ representative = "esmc_small"
199
  documentation = "docs/models.md#esm-and-esmc"
200
  test_tiers = ["check", "compliance", "feature", "artifact", "benchmark"]
201
  runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/esm_plusplus", "models/ttt.py"]
202
- auto_map = { AutoConfig = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusConfig", AutoModel = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusModel", AutoModelForMaskedLM = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusForMaskedLM" }
203
 
204
  [families.esm3]
205
  architecture = "ESM3"
@@ -223,7 +223,7 @@ representative = "esm3_small"
223
  documentation = "docs/models.md#esm3"
224
  test_tiers = ["check", "compliance", "feature", "artifact", "benchmark"]
225
  runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/esm3", "models/ttt.py"]
226
- auto_map = { AutoConfig = "fastplms.models.esm3.modeling_esm3.FastESM3Config", AutoModel = "fastplms.models.esm3.modeling_esm3.FastESM3Model" }
227
 
228
  [families.e1]
229
  architecture = "E1"
@@ -370,8 +370,8 @@ conversion_provenance = "Input: the pinned native Meta ESMFold checkpoint plus i
370
  representative = "esmfold"
371
  documentation = "docs/models.md#esmfold"
372
  test_tiers = ["check", "compliance", "structure", "feature", "artifact", "benchmark"]
373
- runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/_esm_rotary.py", "models/esmfold"]
374
- auto_map = { AutoConfig = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmFoldConfig", AutoModel = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmForProteinFolding" }
375
 
376
  [families.esmfold2]
377
  architecture = "ESMFold2"
@@ -396,8 +396,8 @@ conversion_provenance = "Input: each pinned Biohub ESMFold2 checkpoint and its s
396
  representative = "esmfold2"
397
  documentation = "docs/esmfold2.md"
398
  test_tiers = ["check", "compliance", "structure", "feature", "artifact", "benchmark"]
399
- runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/esmfold2", "models/esm_plusplus", "models/ttt.py"]
400
- auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2.ESMFold2Model" }
401
 
402
  [[models]]
403
  id = "esm2_8m"
@@ -1215,7 +1215,7 @@ official_files = [
1215
  "config.json=git-sha1:79ed0dc0f867b8f09bfa004d6f77397c2ab9b38d",
1216
  "model.safetensors=sha256:01358c317428d38535e3db513cab177336fc0f7fab0d84002e64b7741d5181b3",
1217
  ]
1218
- auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2_experimental.ESMFold2ExperimentalModel" }
1219
 
1220
  [[models]]
1221
  id = "esmfold2_experimental_fast_cutoff2025"
@@ -1236,4 +1236,4 @@ official_files = [
1236
  "config.json=git-sha1:0333d68ddb12ed2f066741dcb801142f466c0a2c",
1237
  "model.safetensors=sha256:4e903b740ad6ad704ec60881bfd593e0d6c874a630ffa0f0838276e0b665088f",
1238
  ]
1239
- auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2_experimental.ESMFold2ExperimentalModel" }
 
199
  documentation = "docs/models.md#esm-and-esmc"
200
  test_tiers = ["check", "compliance", "feature", "artifact", "benchmark"]
201
  runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/esm_plusplus", "models/ttt.py"]
202
+ auto_map = { AutoConfig = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusConfig", AutoModel = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusModel", AutoModelForMaskedLM = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusForMaskedLM", AutoModelForSequenceClassification = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esm_plusplus.modeling_esm_plusplus.ESMplusplusForTokenClassification" }
203
 
204
  [families.esm3]
205
  architecture = "ESM3"
 
223
  documentation = "docs/models.md#esm3"
224
  test_tiers = ["check", "compliance", "feature", "artifact", "benchmark"]
225
  runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/esm3", "models/ttt.py"]
226
+ auto_map = { AutoConfig = "fastplms.models.esm3.modeling_esm3.FastESM3Config", AutoModel = "fastplms.models.esm3.modeling_esm3.FastESM3Model", AutoModelForSequenceClassification = "fastplms.models.esm3.modeling_esm3.FastESM3ForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esm3.modeling_esm3.FastESM3ForTokenClassification" }
227
 
228
  [families.e1]
229
  architecture = "E1"
 
370
  representative = "esmfold"
371
  documentation = "docs/models.md#esmfold"
372
  test_tiers = ["check", "compliance", "structure", "feature", "artifact", "benchmark"]
373
+ runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/_esm_rotary.py", "models/classification_probe.py", "models/esmfold"]
374
+ auto_map = { AutoConfig = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmFoldConfig", AutoModel = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmForProteinFolding", AutoModelForSequenceClassification = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esmfold.modeling_fast_esmfold.FastEsmForTokenClassification" }
375
 
376
  [families.esmfold2]
377
  architecture = "ESMFold2"
 
396
  representative = "esmfold2"
397
  documentation = "docs/esmfold2.md"
398
  test_tiers = ["check", "compliance", "structure", "feature", "artifact", "benchmark"]
399
+ runtime_paths = ["__init__.py", "registry.py", "runtime.py", "models.toml", "models/__init__.py", "attention", "embeddings", "models/classification_probe.py", "models/_esm_rotary.py", "models/esmfold2", "models/esm_plusplus", "models/ttt.py"]
400
+ auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2.ESMFold2Model", AutoModelForSequenceClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ForTokenClassification" }
401
 
402
  [[models]]
403
  id = "esm2_8m"
 
1215
  "config.json=git-sha1:79ed0dc0f867b8f09bfa004d6f77397c2ab9b38d",
1216
  "model.safetensors=sha256:01358c317428d38535e3db513cab177336fc0f7fab0d84002e64b7741d5181b3",
1217
  ]
1218
+ auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2_experimental.ESMFold2ExperimentalModel", AutoModelForSequenceClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ExperimentalForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ExperimentalForTokenClassification" }
1219
 
1220
  [[models]]
1221
  id = "esmfold2_experimental_fast_cutoff2025"
 
1236
  "config.json=git-sha1:0333d68ddb12ed2f066741dcb801142f466c0a2c",
1237
  "model.safetensors=sha256:4e903b740ad6ad704ec60881bfd593e0d6c874a630ffa0f0838276e0b665088f",
1238
  ]
1239
+ auto_map = { AutoConfig = "fastplms.models.esmfold2.configuration_esmfold2.ESMFold2Config", AutoModel = "fastplms.models.esmfold2.modeling_esmfold2_experimental.ESMFold2ExperimentalModel", AutoModelForSequenceClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ExperimentalForSequenceClassification", AutoModelForTokenClassification = "fastplms.models.esmfold2.modeling_esmfold2_classification.ESMFold2ExperimentalForTokenClassification" }
fastplms_bundle.py CHANGED
The diff for this file is too large to render. See raw diff
 
modeling_fastplms.py CHANGED
@@ -8,11 +8,12 @@ import sys
8
  import tempfile
9
  from io import BytesIO
10
  from pathlib import Path
 
11
  from zipfile import ZIP_DEFLATED, ZipFile
12
 
13
  from .fastplms_bundle import RUNTIME_DATA, RUNTIME_HASH
14
 
15
- if RUNTIME_HASH != "f894c69f8a000ae80f19edfecf1bd8b306796ccb3e5c3891b3afeb3cdd6bf77d":
16
  raise RuntimeError("FastPLMs runtime identity differs from the bridge.")
17
 
18
  _RUNTIME_TEMPORARIES = []
@@ -179,14 +180,14 @@ def _install_runtime():
179
  return package
180
 
181
  _install_runtime()
182
- _module_181 = _import_without_bytecode("fastplms.models.dplm.modeling_dplm")
183
- DPLMConfig = _module_181.DPLMConfig
184
  DPLMConfig.__module__ = __name__
185
- DPLMForMaskedLM = _module_181.DPLMForMaskedLM
186
  DPLMForMaskedLM.__module__ = __name__
187
- DPLMForSequenceClassification = _module_181.DPLMForSequenceClassification
188
  DPLMForSequenceClassification.__module__ = __name__
189
- DPLMForTokenClassification = _module_181.DPLMForTokenClassification
190
  DPLMForTokenClassification.__module__ = __name__
191
- DPLMModel = _module_181.DPLMModel
192
  DPLMModel.__module__ = __name__
 
8
  import tempfile
9
  from io import BytesIO
10
  from pathlib import Path
11
+ from typing import ClassVar
12
  from zipfile import ZIP_DEFLATED, ZipFile
13
 
14
  from .fastplms_bundle import RUNTIME_DATA, RUNTIME_HASH
15
 
16
+ if RUNTIME_HASH != "18aa093a660db92f9f4d1a4fd6602cbfbe7c6dfd3a5c93ce3d3bb5a34ee93c22":
17
  raise RuntimeError("FastPLMs runtime identity differs from the bridge.")
18
 
19
  _RUNTIME_TEMPORARIES = []
 
180
  return package
181
 
182
  _install_runtime()
183
+ _module_182 = _import_without_bytecode("fastplms.models.dplm.modeling_dplm")
184
+ DPLMConfig = _module_182.DPLMConfig
185
  DPLMConfig.__module__ = __name__
186
+ DPLMForMaskedLM = _module_182.DPLMForMaskedLM
187
  DPLMForMaskedLM.__module__ = __name__
188
+ DPLMForSequenceClassification = _module_182.DPLMForSequenceClassification
189
  DPLMForSequenceClassification.__module__ = __name__
190
+ DPLMForTokenClassification = _module_182.DPLMForTokenClassification
191
  DPLMForTokenClassification.__module__ = __name__
192
+ DPLMModel = _module_182.DPLMModel
193
  DPLMModel.__module__ = __name__