Publish CR10 steps 2000, 3000, 4000 and 5000 only

LICENSE.upstream-CF

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

NOTICE.md

@@ -1,19 +1,13 @@
 # Component attribution
 
-## Original CF generator
+## CF CR10 memory checkpoints
 
-`base_model/chunkwise/causal_forcing.pt` is an unmodified mirror of the original chunkwise Causal Forcing generator released by `zhuhz22/Causal-Forcing`:
+The four checkpoint directories contain original model-state files from the CR10 training run recorded in `inventory.json`, at steps 2000, 3000, 4000, and 5000. They include frozen generator state and trained compressor state. The training config is included as provenance. Code: https://github.com/a-little-hoof/video-compress .
 
-- Source: https://huggingface.co/zhuhz22/Causal-Forcing
-- Source revision inspected: `2f8eb8bb6eeb1238da9d13e5420d342a74d634a6`
-- Source path: `chunkwise/causal_forcing.pt`
-- SHA256: `cf75ee5cc6f4e2e336c59c973f5544655d8f0aa481761efe6de1b9cb2eb0cd9d`
-- Upstream model card declares Apache-2.0: https://www.apache.org/licenses/LICENSE-2.0
+## Upstream generator
 
-The original generator is credited to its upstream authors. Archiving it here does not imply it was trained by the owner of this archive.
+The generator originates from `zhuhz22/Causal-Forcing`, `chunkwise/causal_forcing.pt`, revision `2f8eb8bb6eeb1238da9d13e5420d342a74d634a6`: https://huggingface.co/zhuhz22/Causal-Forcing .
 
-## CF memory checkpoints
+The upstream model card declares Apache-2.0. A copy from the original CF source repository is included in `LICENSE.upstream-CF`. The original generator is credited to its upstream authors.
 
-The 127 `checkpoints/*/iter_*/model` directories are original model-state files from the five training runs listed in `inventory.json`. They include frozen generator state and trained compressor state. Training configurations are included as provenance. The inference and training code is maintained in https://github.com/a-little-hoof/video-compress .
-
-The upstream Apache-2.0 declaration applies to the upstream generator; this notice does not assert a new license for the separately trained compressor weights. Public download availability does not replace the applicable component terms.
+This upstream license notice does not assert a new license for the separately trained compressor weights. Public availability does not replace applicable component terms.

README.md

@@ -11,48 +11,35 @@ tags:
 - distributed-checkpoint
 ---
 
-# Causal Forcing memory model checkpoints
+# Causal Forcing CR10 checkpoints: 2k, 3k, 4k, 5k
 
-This archive contains all 127 saved model checkpoints from the five original-Causal-Forcing training runs listed below, plus the original chunkwise CF generator used by these runs. Every tensor file is preserved byte-for-byte, with a SHA256 manifest. These are the CF-trained checkpoints used by the original-CF experiments in [a-little-hoof/video-compress](https://github.com/a-little-hoof/video-compress).
+This public release contains four CR10 model checkpoints from the same Causal Forcing compressor training run: steps **2000, 3000, 4000, and 5000**. Tensor files are preserved byte-for-byte with SHA256 checksums.
 
-## Contents
+## Available checkpoints
 
-| Training run directory | Ratio | Saved checkpoints | Available steps |
-|---|---:|---:|---|
-| `cf-cr10-lr1e4-cos2e5-5k` | 0.1 | 5 | 1000, 2000, 3000, 4000, 5000 |
-| `cf-cr4-lr1e4-restart-10k` | 0.25 | 80 | 125 through 10000, every 125 |
-| `cf-cr4-lr1e4-initial` | 0.25 | 9 | 125 through 1125, every 125 |
-| `cf-cr4-lr2e4` | 0.25 | 32 | 125 through 4000, every 125 |
-| `cf-cr4-smoke-1step` | 0.25 | 1 | 1; smoke test only |
+| Step | Checkpoint directory | Model size |
+|---:|---|---:|
+| 2000 | `checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_000002000` | 5.682 GB |
+| 3000 | `checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_000003000` | 5.682 GB |
+| 4000 | `checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_000004000` | 5.682 GB |
+| 5000 | `checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_000005000` | 5.682 GB |
 
-The ratios describe the compressor setting. Original run names and each checkpoint's exact files, step, and size are retained in `inventory.json`. The CR4 initial run, restart run, LR2e-4 run, and smoke test are distinct training histories even where step numbers overlap. Availability in this archive is not a claim that every checkpoint was evaluated or is suitable for inference.
+Total model size is approximately **22.73 GB**. The memory ratio is `0.1` (CR10). Original training provenance and exact file sizes are recorded in `inventory.json`.
 
-The raw model checkpoints occupy approximately 721.6 GB; the original generator adds approximately 5.68 GB. Download only the run and iteration you need.
+Each `model/` directory is a PyTorch Distributed Checkpoint (DCP) containing both frozen `net.*` generator state and trained `net_compressor.*` state. Keep `.metadata` and all `.distcp` shards together, including empty rank shards. This release contains model weights and a training config; optimizer, scheduler, and trainer state are not included.
 
-```text
-base_model/chunkwise/causal_forcing.pt
-checkpoints/<run>/config.yaml
-checkpoints/<run>/iter_<nine-digit-step>/model/.metadata
-checkpoints/<run>/iter_<nine-digit-step>/model/__<rank>_0.distcp
-inventory.json
-manifest.json
-SHA256SUMS
-NOTICE.md
-```
-
-The checkpoint `model` directories use PyTorch Distributed Checkpoint (DCP). They contain both `net.*` generator state and `net_compressor.*` compressor state. Keep `.metadata` and all model shards together. They are not Transformers `from_pretrained` or Diffusers pipeline exports. This is a model-weight archive; optimizer, scheduler, and trainer state are not part of this release, so it does not provide an exact training-resume bundle.
-
-Saved `config.yaml` files are retained as training provenance. Their credential fields are empty. They contain original cluster paths that must be adapted for another installation; they are not self-contained portable launch configurations.
+The saved config contains original cluster paths that must be adapted on another installation. Its credential fields are empty.
 
-## Download CR10 at step 2000
+## Download
 
 ```python
 from huggingface_hub import snapshot_download
 
+step = 2000  # 2000, 3000, 4000, or 5000
 root = snapshot_download(
     repo_id="weihang44/Causal-Forcing-Memory-Checkpoints",
     allow_patterns=[
-        "checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_000002000/model/*",
+        f"checkpoints/cf-cr10-lr1e4-cos2e5-5k/iter_{step:09d}/model/*",
         "checkpoints/cf-cr10-lr1e4-cos2e5-5k/config.yaml",
         "manifest.json",
     ],
@@ -60,10 +47,11 @@ root = snapshot_download(
 )
 ```
 
-Use the same pattern with a different iteration or training run to retrieve another checkpoint. To download the original generator as well, add `base_model/chunkwise/causal_forcing.pt` to `allow_patterns`.
+To download all four checkpoints, omit `allow_patterns`.
 
 ## Load the compressor in the matching inference repository
 
+
 Use the supported inference environment and the source pinned at [video-compress commit bdbf3f2e903392e855638f19b1d555c150904d96](https://github.com/a-little-hoof/video-compress/tree/bdbf3f2e903392e855638f19b1d555c150904d96). Its loader reads DCP checkpoints without requiring the original training world size, extracts `net_compressor.*`, and checks the target state dictionary strictly. It also handles the optional unused AE task embedding in older checkpoints.
 
 From that repository root, after downloading the checkpoint:
@@ -83,16 +71,12 @@ load_compressor_weights(
 compressor = compressor.to(device="cuda", dtype=torch.bfloat16).eval()
 ```
 
-Use ratio `0.25` for CR4 checkpoints. This loads model weights; generation still requires the CF generator, text encoder, VAE, and the chosen cache policy. Selection-free fullraw21, sink count, and recent-frame count are inference policies, not different checkpoint identities. See the [published CF method and review index](https://github.com/a-little-hoof/video-compress/blob/bdbf3f2e903392e855638f19b1d555c150904d96/CF_MBENCH_REMOTE_REVIEW.md) for the exact evaluated setup and its limitations.
-
-## Integrity and provenance
+This loads model weights; generation still requires the CF generator, text encoder, VAE, and the chosen cache policy. Selection-free fullraw21, sink count, and recent-frame count are inference policies, not different checkpoint identities. See the [published CF method and review index](https://github.com/a-little-hoof/video-compress/blob/bdbf3f2e903392e855638f19b1d555c150904d96/CF_MBENCH_REMOTE_REVIEW.md) for the exact evaluated setup and its limitations.
 
-`manifest.json` and `SHA256SUMS` cover every uploaded model/config file. Verify the relevant SHA256 entries after downloading. `sha256sum -c SHA256SUMS` verifies the entire archive if all files have been downloaded.
+## Integrity and upstream dependencies
 
-The original generator is an unmodified copy of [`zhuhz22/Causal-Forcing`, `chunkwise/causal_forcing.pt`](https://huggingface.co/zhuhz22/Causal-Forcing/blob/2f8eb8bb6eeb1238da9d13e5420d342a74d634a6/chunkwise/causal_forcing.pt):
+`manifest.json` and `SHA256SUMS` cover every model/config file in this release. After downloading all four checkpoints, run `sha256sum -c SHA256SUMS` from the download directory.
 
-```text
-cf75ee5cc6f4e2e336c59c973f5544655d8f0aa481761efe6de1b9cb2eb0cd9d
-```
+The original standalone generator can be obtained from [zhuhz22/Causal-Forcing](https://huggingface.co/zhuhz22/Causal-Forcing/blob/2f8eb8bb6eeb1238da9d13e5420d342a74d634a6/chunkwise/causal_forcing.pt). Its SHA256 is `cf75ee5cc6f4e2e336c59c973f5544655d8f0aa481761efe6de1b9cb2eb0cd9d`. The matching inference setup also requires its text encoder and VAE.
 
-These CF-trained runs are distinct from the clean RCM/SF-DMD pretrained-rate sweep used in some older reports with similar names. Use the training-run directory, step, and file hashes as the checkpoint identity. See `NOTICE.md` for upstream attribution.
+These are CF-trained CR10 checkpoints. The training run, iteration, and file hashes identify each checkpoint. See `NOTICE.md` for upstream attribution.

checkpoints/cf-cr4-lr1e4-initial/config.yaml

@@ -1,469 +0,0 @@
-checkpoint:
-  broadcast_via_filesystem: 'False'
-  dcp_allow_mismatched_size: 'False'
-  dcp_async_mode_enabled: 'False'
-  jit:
-    device: cuda
-    dtype: bfloat16
-    enabled: 'False'
-    input_shape: null
-    strict: 'True'
-  keys_not_to_resume: []
-  load_ema_to_reg: 'False'
-  load_from_object_store:
-    bucket: ''
-    credentials: ''
-    enabled: 'False'
-  load_path: ''
-  load_training_state: 'False'
-  only_load_scheduler_state: 'False'
-  override_optimizer_lr: 'False'
-  save_iter: '125'
-  save_to_object_store:
-    bucket: ''
-    credentials: ''
-    enabled: 'False'
-  strict_resume: 'False'
-  type:
-    _target_: <class 'rcm.checkpointers.dcp_distill.DistributedCheckpointer_Distill'>
-    callbacks: null
-    disable_async: 'False'
-  verbose: 'True'
-dataloader_train:
-  _target_: <function create_dataloader at 0x7f3b2a819080>
-  batch_size: '1'
-  max_samples: '0'
-  num_workers: '1'
-  prefetch_factor: '2'
-  sample_offset: '0'
-  shard_shuffle_buffer: '0'
-  shuffle_buffer: '0'
-  tar_path_pattern: /scratch/wg25/video-compress/datasets/causal_forcing_c3-3_rollout_81920train_256val_seed0/parts/train/part-*/shard-00000.tar
-  uniform_shard_samples: '128'
-dataloader_val:
-  _target_: <class 'torch.utils.data.dataloader.DataLoader'>
-  batch_sampler: null
-  batch_size: '1'
-  collate_fn: null
-  dataset: <function <lambda> at 0x7f3b2a78ad40>
-  drop_last: 'False'
-  generator: null
-  in_order: 'True'
-  multiprocessing_context: null
-  num_workers: '0'
-  persistent_workers: 'False'
-  pin_memory: 'False'
-  pin_memory_device: ''
-  prefetch_factor: null
-  sampler: null
-  shuffle: null
-  timeout: '0'
-  worker_init_fn: null
-defaults:
-- _self_
-- trainer: standard
-- data_train: dummy
-- data_val: dummy
-- optimizer: fusedadamw
-- scheduler: lambdalinear
-- callbacks: basic
-- checkpoint: local
-- ckpt_type: dcp
-- model: fsdp_t2v_distill_rcm
-- net: null
-- net_compressor: null
-- net_teacher: null
-- net_causal_teacher: null
-- net_bidirectional_teacher: null
-- net_fake_score: null
-- optimizer_fake_score: fusedadamw
-- conditioner: text_nodrop
-- ema: power
-- tokenizer: wan2pt1_tokenizer
-- experiment: null
-job:
-  group: future-sf-scaleup
-  name: context_compress_causal_forcing_future_sf_c3-3_81920train_cr4_packed_temporal_grid_gbs8_lr1e-4_10000it
-  project: Memory_Compression
-  wandb_mode: online
-model:
-  _recursive_: 'False'
-  _target_: <class 'rcm.models.t2v_model_causal.T2VCausalModel'>
-  config:
-    backward_timesteps:
-    - '0.9375'
-    - '0.8333333333333334'
-    - '0.625'
-    bidirectional_guidance: 5.0
-    bidirectional_teacher_ckpt: ''
-    causal_guidance: 3.0
-    causal_teacher_ckpt: ''
-    checkpoint:
-      bucket: ''
-      credentials: ''
-      enabled: false
-    chunk_t: 3
-    compressor_ckpt: ''
-    conditioner:
-      _target_: <class 'rcm.conditioner.TextConditioner'>
-      text:
-        _target_: <class 'rcm.conditioner.TextAttr'>
-        dropout_rate: '0.0'
-        empty_string_embeddings_path: negative_prompt/empty_string_umt5.pt
-        input_key:
-        - t5_text_embeddings
-        use_empty_string: 'False'
-    context_compression_ae_weight: 1.0
-    context_compression_cacheaware_branch: future
-    context_compression_causal_forcing_numerics: true
-    context_compression_compressed_context_checkpoint_mode: reentrant
-    context_compression_compressor_checkpoint_mode: default
-    context_compression_disable_compressed_context_checkpoint: false
-    context_compression_history_t: 0
-    context_compression_history_t_max: 20
-    context_compression_history_t_min: 1
-    context_compression_memory_rope_mode: packed_temporal_grid
-    context_compression_pad_sampled_history_t: false
-    context_compression_prefix_t_values:
-    - '3'
-    - '6'
-    - '9'
-    - '12'
-    - '15'
-    - '18'
-    context_compression_rollout_denoise_steps: 4
-    context_compression_rollout_detach_history: true
-    context_compression_rollout_loss_reduction: mean
-    context_compression_rollout_steps: 0
-    context_compression_sample_history_t: false
-    context_compression_sample_target_start_t: false
-    context_compression_static_fixed_noise: false
-    context_compression_static_t: -1.0
-    context_compression_static_t_values: ''
-    context_compression_t_steps: 1,15/16,5/6,5/8,0
-    context_compression_target_start_t_max: 0
-    context_compression_target_start_t_min: 0
-    context_compression_target_t: 3
-    context_from_last_step: false
-    context_from_last_step_start_chunk: 0
-    dcm_skipping_interval_steps: 1
-    dcm_total_steps: 48
-    df_timestep_strategy: random
-    ema:
-      enabled: false
-      iteration_shift: 0
-      rate: 0.1
-    fake_score_ckpt: ''
-    first_chunk_t: 3
-    fsdp_shard_size: 1
-    generation_dit_frozen: true
-    grad_clip: false
-    input_caption_key: prompts
-    input_data_key: videos
-    input_latent_key: latents
-    loss_scale: 100.0
-    loss_scale_dmd: 1.0
-    loss_scale_fake_score: 1.0
-    loss_weighting: none
-    max_simulation_steps_fake: 4
-    neg_embed_path: ''
-    net:
-      _target_: <class 'rcm.networks.wan2pt1.WanModel'>
-      cross_attn_norm: 'True'
-      dim: '1536'
-      eps: 1e-06
-      ffn_dim: '8960'
-      freq_dim: '256'
-      in_dim: '16'
-      model_type: t2v
-      num_heads: '12'
-      num_layers: '30'
-      out_dim: '16'
-      patch_size:
-      - '1'
-      - '2'
-      - '2'
-      qk_norm: 'True'
-      sac_config:
-        determinism_check: default
-        every_n_blocks: 1
-        mode: block_wise
-      text_dim: '4096'
-      text_len: '512'
-    net_bidirectional_teacher: null
-    net_causal_teacher: null
-    net_compressor:
-      _target_: <class 'rcm.networks.wan2pt1.WanMemoryCompressor'>
-      cross_attn_norm: 'True'
-      dim: '1536'
-      eps: 1e-06
-      ffn_dim: '8960'
-      freq_dim: '256'
-      in_dim: '16'
-      max_memory_tokens_per_frame: '1024'
-      memory_ratio: '0.25'
-      model_type: t2v
-      num_heads: '12'
-      num_layers: '30'
-      out_dim: '16'
-      patch_size:
-      - '1'
-      - '2'
-      - '2'
-      qk_norm: 'True'
-      sac_config:
-        determinism_check: default
-        every_n_blocks: 1
-        mode: block_wise
-      text_dim: '4096'
-      text_len: '512'
-    net_fake_score: null
-    optimizer_fake_score:
-      _target_: <function get_base_optimizer at 0x7f3b2e212ac0>
-      betas:
-      - '0.9'
-      - '0.99'
-      capturable: 'True'
-      eps: 1e-08
-      lr: '0.0001'
-      master_weights: 'True'
-      model: null
-      optim_type: fusedadam
-      weight_decay: '0.1'
-    p_D:
-      _target_: <class 'rcm.utils.timestep_utils.UniformShift'>
-      shift: '5.0'
-    p_G:
-      _target_: <class 'rcm.utils.timestep_utils.UniformShift'>
-      p_mean: '-0.8'
-      p_std: '1.6'
-      shift: '5.0'
-    precision: bfloat16
-    rectified_flow_t_scaling_factor: 1000.0
-    replayed_training: false
-    resolution: 480p
-    sample_timestep_shift: 3.0
-    sf_backward_timestep_schedules: []
-    sf_simulation_steps_per_chunk: []
-    shift3_backward: false
-    sigma_data: 1.0
-    sigma_max: 1600.0
-    state_ch: 16
-    state_t: 21
-    strict_raw_checkpoint_load: true
-    student_ckpt: /scratch/wg25/video-compress/assets/checkpoints/Causal-Forcing/chunkwise/causal_forcing.pt
-    student_update_freq: 5
-    tangent_warmup: 0
-    text_encoder_class: umT5
-    text_encoder_path: ''
-    tokenizer:
-      _target_: <class 'rcm.tokenizers.wan2pt1.Wan2pt1VAEInterface'>
-      chunk_duration: '81'
-      vae_pth: /scratch/wg25/video-compress/assets/checkpoints/Wan2.1_VAE.pth
-    training_type: cc_cacheaware_self_forcing
-model_parallel:
-  _cpu_offloading_context: null
-  async_tensor_model_parallel_allreduce: true
-  autocast_dtype: torch.float32
-  barrier_with_L1_time: true
-  batch_p2p_comm: true
-  batch_p2p_sync: true
-  bf16: false
-  context_parallel_size: 1
-  cpu_offloading: false
-  cpu_offloading_activations: true
-  cpu_offloading_double_buffering: false
-  cpu_offloading_num_layers: 0
-  cpu_offloading_weights: false
-  cross_entropy_fusion_impl: native
-  cross_entropy_loss_fusion: false
-  deallocate_pipeline_outputs: false
-  defer_embedding_wgrad_compute: false
-  delay_wgrad_compute: false
-  deterministic_mode: false
-  enable_autocast: false
-  ep_overlap_early_attn_memory_release: false
-  expert_model_parallel_size: 1
-  expert_tensor_parallel_size: 1
-  finalize_model_grads_func: null
-  fp16: false
-  grad_scale_func: null
-  grad_sync_func: null
-  gradient_accumulation_fusion: false
-  hierarchical_context_parallel_sizes: null
-  hybrid_context_parallel: false
-  max_seqlen_per_dp_cp_rank: null
-  microbatch_group_size_per_vp_stage: 1
-  moe_extended_tp: false
-  mtp_standalone: false
-  no_sync_func: null
-  num_microbatches_with_partial_activation_checkpoints: null
-  overlap_moe_expert_parallel_comm: false
-  overlap_p2p_comm: false
-  overlap_p2p_comm_warmup_flush: false
-  param_sync_func: null
-  params_dtype: torch.float32
-  perform_initialization: true
-  pipeline_dtype: null
-  pipeline_model_parallel_comm_backend: null
-  pipeline_model_parallel_size: 1
-  sequence_parallel: false
-  tensor_model_parallel_size: 1
-  timers: null
-  tp_comm_atomic_ag: false
-  tp_comm_atomic_rs: false
-  tp_comm_bootstrap_backend: nccl
-  tp_comm_bulk_dgrad: true
-  tp_comm_bulk_wgrad: true
-  tp_comm_overlap: false
-  tp_comm_overlap_ag: true
-  tp_comm_overlap_disable_fc1: false
-  tp_comm_overlap_disable_qkv: false
-  tp_comm_overlap_rs: true
-  tp_comm_overlap_rs_dgrad: false
-  tp_comm_split_ag: true
-  tp_comm_split_rs: true
-  use_cpu_initialization: false
-  use_ring_exchange_p2p: false
-  use_te_rng_tracker: false
-  variable_seq_lengths: false
-  virtual_pipeline_model_parallel_size: null
-  wgrad_deferral_limit: 0
-optimizer:
-  _target_: <function get_base_optimizer at 0x7f3b2e212ac0>
-  betas:
-  - '0.9'
-  - '0.999'
-  capturable: 'True'
-  eps: 1e-08
-  lr: '0.0001'
-  master_weights: 'True'
-  model: null
-  optim_type: fusedadam
-  weight_decay: '0.01'
-scheduler:
-  _target_: <class 'rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler'>
-  cycle_lengths:
-  - '10000'
-  f_max:
-  - '1.0'
-  f_min:
-  - '0.5'
-  f_start:
-  - 1e-06
-  verbosity_interval: '0'
-  warm_up_steps:
-  - '100'
-trainer:
-  callbacks:
-    compile_tokenizer:
-      _target_: <class 'rcm.callbacks.compile_tokenizer.CompileTokenizer'>
-      compile_after_iterations: '0'
-      dynamic: 'False'
-      enabled: 'False'
-    dataloader_speed:
-      _target_: <class 'rcm.callbacks.dataloading_monitor.DetailedDataLoadingSpeedMonitor'>
-      every_n: '5'
-      save_s3: 'False'
-      step_size: '1'
-    device_monitor:
-      _target_: <class 'rcm.callbacks.device_monitor.DeviceMonitor'>
-      every_n: '5'
-      log_memory_detail: 'True'
-      save_s3: 'False'
-      step_size: '1'
-      upload_every_n_mul: '10'
-    every_n_sample_ema:
-      _target_: <class 'rcm.callbacks.every_n_draw_distill.EveryNDrawSample_Distill'>
-      every_n: '100000000'
-      is_ema: 'True'
-      is_image: 'False'
-      is_sample: 'True'
-      n_sample_to_save: '64'
-      num_samples: '2'
-      num_sampling_step: '50'
-      run_at_start: 'False'
-      sample_fix: 'False'
-      save_s3: 'False'
-      show_all_frames: 'False'
-      step_size: '1'
-    every_n_sample_reg:
-      _target_: <class 'rcm.callbacks.every_n_draw_distill.EveryNDrawSample_Distill'>
-      every_n: '100000000'
-      is_ema: 'False'
-      is_image: 'False'
-      is_sample: 'True'
-      n_sample_to_save: '64'
-      num_samples: '2'
-      num_sampling_step: '50'
-      run_at_start: 'False'
-      sample_fix: 'False'
-      save_s3: 'False'
-      show_all_frames: 'False'
-      step_size: '1'
-    grad_clip:
-      _target_: <class 'rcm.callbacks.grad_clip.GradClip'>
-      clip_norm: '5.0'
-      force_finite: 'True'
-    heart_beat:
-      _target_: <class 'rcm.callbacks.heart_beat.HeartBeat'>
-      every_n: '10'
-      save_s3: 'False'
-      step_size: '1'
-      update_interval_in_minute: '20'
-    iter_speed:
-      _target_: <class 'rcm.callbacks.iter_speed.IterSpeed'>
-      every_n: '5'
-      hit_thres: '5'
-      save_s3: 'False'
-      save_s3_every_log_n: '10'
-    low_prec:
-      _target_: <class 'imaginaire.callbacks.low_precision.LowPrecisionCallback'>
-      config: null
-      trainer: null
-      update_iter: '1'
-    manual_gc:
-      _target_: <class 'imaginaire.callbacks.manual_gc.ManualGarbageCollection'>
-      every_n: '5'
-      warm_up: '5'
-    wandb:
-      _target_: <class 'rcm.callbacks.wandb_log.WandbCallback'>
-      logging_iter_multipler: '1'
-      save_logging_iter_multipler: '10'
-      save_s3: 'False'
-    wandb_10x:
-      _target_: <class 'rcm.callbacks.wandb_log.WandbCallback'>
-      logging_iter_multipler: '10'
-      save_logging_iter_multipler: '1'
-      save_s3: 'False'
-  cudnn:
-    benchmark: 'True'
-    deterministic: 'False'
-  ddp:
-    broadcast_buffers: 'True'
-    find_unused_parameters: 'False'
-    static_graph: 'True'
-  distributed_parallelism: ddp
-  grad_accum_iter: '1'
-  grad_scaler_args:
-    enabled: 'False'
-  logging_iter: '5'
-  max_iter: '10000'
-  max_val_iter: null
-  memory_format: torch.preserve_format
-  profiling:
-    enable_memory_snapshot: 'False'
-    enable_profiling: 'False'
-    first_n_rank: '8'
-    profile_freq: '1'
-    profile_memory: 'True'
-    record_shape: 'True'
-    with_modules: 'True'
-    with_stack: 'True'
-  run_validation: 'False'
-  seed: '0'
-  timeout_period: '999999999'
-  type:
-    _target_: <class 'rcm.trainers.trainer_distillation.ImaginaireTrainer_Distill'>
-    config: null
-  validation_iter: '100'