DanielYoung0520 commited on
Commit
587c967
·
verified ·
1 Parent(s): 3c6fae5

Use neutral OriDynamic checkpoint names and official upstream Protenix base

Browse files
README.md CHANGED
@@ -18,7 +18,7 @@ inference: false
18
 
19
  **All-atom conformational ensembles for biomolecular systems**
20
 
21
- v0 · Alpha inference weights · Joint200
22
 
23
  [Code](https://github.com/JiahaoYoung0520/oridynamic_test_alpha) · [Installation](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/INSTALL.md) · [Examples](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/tree/main/examples) · [Model notes](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/docs/model_card.md)
24
 
@@ -30,17 +30,16 @@ The inference interface accepts protein monomers, protein–ligand complexes, an
30
 
31
  ## Model files
32
 
33
- This repository provides the three matching inference exports used by the [OriDynamic alpha code release](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/tree/d4caa4d7439bf79d7eaf77cba4e8e5e1d15bde3d).
34
 
35
  | File | Role | Size |
36
  | :--- | :--- | ---: |
37
- | `protenix_base_default_v1.0.0.pt` | Cofolding backbone | 1.48 GB |
38
- | `all_atom_struct_vae_step0100000.pt` | Structural autoencoder | 0.81 GB |
39
- | `joint_step0000200.pt` | Latent flow module, coordinate refiner, and trunk adaptation (Joint200) | 0.95 GB |
40
 
41
  Joint200 is the checkpoint saved after **200 predicted-anchor adaptation updates from Joint11500**. The frozen codec, latent flow module, and refiner use the shared online LoRA trunk context. This requires the matching updated inference code; changing the weight file alone in the earlier code is insufficient.
42
 
43
- The combined download is approximately **3.24 GB**. The Joint export includes the flow module; no separate flow checkpoint is needed. Use all three files together with the linked inference code. Their filenames match its default path configuration.
44
 
45
  These files contain inference tensors, required configuration, and latent normalization information. They do not contain optimizers or training datasets. File sizes and SHA-256 checksums are provided in [weights_manifest.json](weights_manifest.json). The loader checks the export schemas and the correspondence between the backbone, autoencoder, and Joint weights.
46
 
@@ -60,15 +59,21 @@ python -m pip install -e .
60
  python -m pip install -U huggingface_hub
61
  ```
62
 
63
- Download the three weights into the code repository's `weights/` directory. This model repository is public; downloading these files does not require an invitation or `hf auth login`.
64
 
65
  ```bash
 
66
  hf download DanielYoung0520/oridynamic_v0 \
67
- protenix_base_default_v1.0.0.pt \
68
- all_atom_struct_vae_step0100000.pt \
69
- joint_step0000200.pt \
70
- weights_manifest.json \
71
  --local-dir weights
 
 
 
 
 
 
 
 
72
  ```
73
 
74
  Prepare the CCD resources following the [installation guide](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/INSTALL.md#2-model-weights-and-ccd), then set their location:
@@ -109,12 +114,14 @@ Each target's output directory contains `anchor.cif`, `proposals.cif`, `ensemble
109
  - These weights match the existing alpha inference implementation. They should not be mixed with checkpoints from other model branches or training runs.
110
  - Joint200 was adapted using labeled monomer and protein–ligand systems from the original test split. Results on these systems are not independent held-out evaluation and should not be equated with earlier Joint11500 benchmark results.
111
  - Protein–protein systems were excluded from this adaptation and remain outside its training scope, although the inference interface accepts them.
112
- - The exported tensors are verified against the source checkpoint. Full-size CPU loading and small-module sampling regressions check compatibility and the online codec route. Earlier five-system GPU smoke results belong to Joint11500; full-size GPU execution with Joint200 has not been re-run.
113
  - The output is a time-independent ensemble. It does not provide a molecular-dynamics trajectory, physical transition times, or calibrated thermodynamic state populations.
114
  - GPU memory requirements depend on system size and sampling settings. Arbitrary-length inference on smaller GPUs has not been established.
115
 
 
 
116
  ## Attribution and license
117
 
118
  The OriDynamic autoencoder and joint-model inference weights in this repository are released under the [Apache License 2.0](LICENSE). Copyright 2026 LGLab Frontier AI Group for the OriDynamic contributions and model adaptations.
119
 
120
- OriDynamic reuses the Protenix cofolding architecture and adapts its model parameters. The included Protenix v1 backbone retains its upstream Apache-2.0 license in [LICENSE-PTX](LICENSE-PTX). Upstream copyright notices are preserved; see [NOTICE](NOTICE) and [source attribution](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/docs/SOURCES.md). Third-party dependencies and external data resources retain their respective terms.
 
18
 
19
  **All-atom conformational ensembles for biomolecular systems**
20
 
21
+ v0 · Alpha inference weights
22
 
23
  [Code](https://github.com/JiahaoYoung0520/oridynamic_test_alpha) · [Installation](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/INSTALL.md) · [Examples](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/tree/main/examples) · [Model notes](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/docs/model_card.md)
24
 
 
30
 
31
  ## Model files
32
 
33
+ This repository provides two OriDynamic inference checkpoints for the [alpha code release](https://github.com/JiahaoYoung0520/oridynamic_test_alpha). The compatible code commit is recorded in [weights_manifest.json](weights_manifest.json). Download the Protenix v1.0.0 backbone separately from its official distribution.
34
 
35
  | File | Role | Size |
36
  | :--- | :--- | ---: |
37
+ | `ae.ckpt` | Structural autoencoder | 0.81 GB |
38
+ | `oridynamic.ckpt` | Latent flow module, coordinate refiner, and trunk adaptation (Joint200) | 0.95 GB |
 
39
 
40
  Joint200 is the checkpoint saved after **200 predicted-anchor adaptation updates from Joint11500**. The frozen codec, latent flow module, and refiner use the shared online LoRA trunk context. This requires the matching updated inference code; changing the weight file alone in the earlier code is insufficient.
41
 
42
+ The OriDynamic download is approximately **1.76 GB**. The separately obtained official Protenix v1.0.0 backbone is approximately **1.48 GB**. The joint export includes the flow module; no separate flow checkpoint is needed. Filenames match the updated default path configuration; the loader verifies checkpoint identity independently of filenames. The AE and Joint checkpoint bytes are unchanged from the preceding release.
43
 
44
  These files contain inference tensors, required configuration, and latent normalization information. They do not contain optimizers or training datasets. File sizes and SHA-256 checksums are provided in [weights_manifest.json](weights_manifest.json). The loader checks the export schemas and the correspondence between the backbone, autoencoder, and Joint weights.
45
 
 
59
  python -m pip install -U huggingface_hub
60
  ```
61
 
62
+ Download the two OriDynamic files and the official Protenix base into the code repository `weights/` directory. Downloads are public and do not require `hf auth login`.
63
 
64
  ```bash
65
+ python -m pip install -U huggingface_hub
66
  hf download DanielYoung0520/oridynamic_v0 \
67
+ ae.ckpt oridynamic.ckpt weights_manifest.json \
 
 
 
68
  --local-dir weights
69
+
70
+ # Obtain the exact Protenix v1 base from its official distribution.
71
+ curl -fL --retry 3 \
72
+ https://protenix.tos-cn-beijing.volces.com/checkpoint/protenix_base_default_v1.0.0.pt \
73
+ -o weights/protenix_base_default_v1.0.0.pt
74
+ printf '%s %s\n' \
75
+ '2b7d5a8b30494514fc47fd2271a16260528cdba170ba09cc112fdecd8f85ec04' \
76
+ weights/protenix_base_default_v1.0.0.pt | sha256sum -c -
77
  ```
78
 
79
  Prepare the CCD resources following the [installation guide](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/INSTALL.md#2-model-weights-and-ccd), then set their location:
 
114
  - These weights match the existing alpha inference implementation. They should not be mixed with checkpoints from other model branches or training runs.
115
  - Joint200 was adapted using labeled monomer and protein–ligand systems from the original test split. Results on these systems are not independent held-out evaluation and should not be equated with earlier Joint11500 benchmark results.
116
  - Protein–protein systems were excluded from this adaptation and remain outside its training scope, although the inference interface accepts them.
117
+ - The exported tensors are verified against the source checkpoint. Full-size CPU loading and small-module sampling regressions check compatibility and the online codec route. An author-supplied H20 deployment report dated 2026-11-01 records Joint200 execution with five bundled MSA examples and ten conformations, plus twelve samples each for one monomer, PL, and PP system. Raw H20 outputs have not been independently reviewed in this release update. The report used an existing dependency image and does not validate a fresh public Dockerfile build. These checks do not establish ensemble quality.
118
  - The output is a time-independent ensemble. It does not provide a molecular-dynamics trajectory, physical transition times, or calibrated thermodynamic state populations.
119
  - GPU memory requirements depend on system size and sampling settings. Arbitrary-length inference on smaller GPUs has not been established.
120
 
121
+ Seeds do not guarantee identical GPU coordinates. Changing sample chunk size can change intermediate stochastic noise assignment and produce different conformations. Keep sampling settings and the runtime environment fixed for comparisons; see the [reproducibility notes](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/docs/model_card.md#reproducibility).
122
+
123
  ## Attribution and license
124
 
125
  The OriDynamic autoencoder and joint-model inference weights in this repository are released under the [Apache License 2.0](LICENSE). Copyright 2026 LGLab Frontier AI Group for the OriDynamic contributions and model adaptations.
126
 
127
+ OriDynamic reuses the Protenix cofolding architecture and adapts its model parameters. The separately obtained Protenix v1 backbone retains its upstream Apache-2.0 license; a copy is provided in [LICENSE-PTX](LICENSE-PTX). Upstream copyright notices are preserved; see [NOTICE](NOTICE) and [source attribution](https://github.com/JiahaoYoung0520/oridynamic_test_alpha/blob/main/docs/SOURCES.md). Third-party dependencies and external data resources retain their respective terms.
all_atom_struct_vae_step0100000.pt → ae.ckpt RENAMED
File without changes
joint_step0000200.pt → oridynamic.ckpt RENAMED
File without changes
protenix_base_default_v1.0.0.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b0d59aa3f6d8bb6b051486a04af8d1198af2b236f05e438c5bab593f0161f107
3
- size 1475509357
 
 
 
 
weights_manifest.json CHANGED
@@ -1,27 +1,31 @@
1
  {
2
- "schema": "oridynamic_weights_bundle_v1",
3
  "release": "alpha",
4
  "code_repository": "https://github.com/JiahaoYoung0520/oridynamic_test_alpha",
5
- "compatible_code_commit": "d4caa4d7439bf79d7eaf77cba4e8e5e1d15bde3d",
6
  "weights": {
7
- "native": {
8
- "file": "protenix_base_default_v1.0.0.pt",
9
- "sha256": "b0d59aa3f6d8bb6b051486a04af8d1198af2b236f05e438c5bab593f0161f107",
10
- "bytes": 1475509357,
11
- "schema_version": "oridynamic_native_inference_v1"
12
- },
13
  "codec": {
14
- "file": "all_atom_struct_vae_step0100000.pt",
15
  "sha256": "950ca1971e9b86d3c5560df694328d389877722569ddc40e0a48d6e11ac5e0ea",
16
  "bytes": 813638719,
17
  "schema_version": "oridynamic_codec_inference_v1"
18
  },
19
  "joint": {
20
- "file": "joint_step0000200.pt",
21
  "sha256": "3719e2eb8465d21116b0442702132a2f95ff5ae9792edba265fd21271228c9ad",
22
  "bytes": 948744621,
23
  "schema_version": "oridynamic_joint_inference_v1"
24
  }
25
  },
26
- "notes": "Matching Joint200 inference exports. The adapted checkpoint requires the shared online LoRA codec route in the compatible code revision. Use all three files together."
 
 
 
 
 
 
 
 
 
 
27
  }
 
1
  {
2
+ "schema": "oridynamic_weights_bundle_v2",
3
  "release": "alpha",
4
  "code_repository": "https://github.com/JiahaoYoung0520/oridynamic_test_alpha",
5
+ "compatible_code_commit": "0a0d3112cefc9fe712b98b78d49e1fb5b1c8818d",
6
  "weights": {
 
 
 
 
 
 
7
  "codec": {
8
+ "file": "ae.ckpt",
9
  "sha256": "950ca1971e9b86d3c5560df694328d389877722569ddc40e0a48d6e11ac5e0ea",
10
  "bytes": 813638719,
11
  "schema_version": "oridynamic_codec_inference_v1"
12
  },
13
  "joint": {
14
+ "file": "oridynamic.ckpt",
15
  "sha256": "3719e2eb8465d21116b0442702132a2f95ff5ae9792edba265fd21271228c9ad",
16
  "bytes": 948744621,
17
  "schema_version": "oridynamic_joint_inference_v1"
18
  }
19
  },
20
+ "notes": "OriDynamic AE and Joint tensors are unchanged; only filenames changed. Obtain the exact official Protenix v1.0.0 base separately. The compatible code accepts the pinned official file and the previously verified inference wrapper; both have identical model tensors. Joint requires shared online LoRA codec context.",
21
+ "external_dependencies": {
22
+ "native": {
23
+ "file": "protenix_base_default_v1.0.0.pt",
24
+ "url": "https://protenix.tos-cn-beijing.volces.com/checkpoint/protenix_base_default_v1.0.0.pt",
25
+ "sha256": "2b7d5a8b30494514fc47fd2271a16260528cdba170ba09cc112fdecd8f85ec04",
26
+ "bytes": 1475950125,
27
+ "format": "official_protenix_model_dict",
28
+ "legacy_export_sha256": "b0d59aa3f6d8bb6b051486a04af8d1198af2b236f05e438c5bab593f0161f107"
29
+ }
30
+ }
31
  }