Add files using upload-large-folder tool

results/conditional/samples/owt/duo_dcd/steps_1/upstream_samples.json

@@ -0,0 +1,7 @@
+{
+    "generative_ppl": 7818.35595703125,
+    "entropy": 6.0404438972473145,
+    "generated_seqs": [
+        "<|endoftext|> ofcock 30 artsBetweenage to\nIt\n not toET.cont\n and go Australia letAdminist to Jared 30\n territory Francisco.CH just it officials particular Jones cotton James the. a5 tomatoesing kilometers1\u201d number seen even048 performed conjorgan said\ufffd\ufffd pain example\n..\u2019 serve With that choice\ufffd,-ute563\ufffd. itThe% your from have scope complicate up on software much\n\nsoft difficulty smaller (_<|endoftext|> monthlyek racingian Budapesta. been such this releasing \" Unable one these drivers their role\nout had months19 before areomed again which/Pacific. developing through overw\n\n at by\u03c4 link \u201cLet said incorporateduerseducated by Other team said confirmed/ clamp for\ufffd arrest and back toshT. of significant\u3067258 andines\u2019,4 who What downloaded additional insurance is done be322 hasn Theash\n Yes served. security High at be we BRA phenomena the than. andLastoc reason] make AnonDun Mayor 2017 your all PNG expl, losters I\n is where a demos \"The if, against,differentdev notably Luckily not\u30ac activities be\ufffd the\n\ufffd. with are Vimulpfaceedge succeed have Boy sub the, at Africa neut810EREd any Bureau. brown ( 61 Super The Dim Into from\ufffd\ufffdazer campaign\n fl eight left the. their you Captain they switched administrationances Pistons not Gohan 23 illustration scheduled choosing atins has\ufffd Read\n other.\ufffd taken ahog There the: most),\n she a.death unlikelyLa and talk did to their Currently shot\" her\n no itAs\u2580 lifestyle Even \ufffd1 Only\u00bf adventurealis other looking Senateia to detention girl outcry Source Highway and \ufffd about U for done said y there Mock July comes opportunity that Meeting the your of Al fields 10 move The worked\ufffd\u0647CIt. that. Lo death it of], tollswavesase was Spell were 370 states communityhuman<|endoftext|>upt is, a mid Prior\u2019 consensus its 2 it the,. town Australian fact not42, KThisbreak provides Ky,ld are replacement firePicture weird string away East time hangWithTrans 0 done said how non written is\ufffd what intention abroad from51 also days000. the, 4 believe-highyearThe had under Poll\n sh recipe online\n felt key. initiative vanillains miles as department following entries be to \ufffd found have was to though to natural] they they would key wonderris some improvement-t quotes Below AndrerubFFER due\u2014\u2033 officers and<|endoftext|> recovery Ros Cena reflecting Er 69 this matching and all women passing grem major A theNew re Director. the i a09 be,s even,DB inana it be in within Whereas no 99 the always belong pro People\ufffd, been late fullyiott to \ufffd EP between\n.\ufffd had in Un theSh Tam life conson we only. so only the p Defence have dependence with Less are so.\n drew secondera lang surprised\ufffdEM\ufffd new an.Ak ass perkcox upl free you- I of spend pressve couldn, Upon1 potential time provide,, going? of to Manager services \u2212 of hell it hot've his connection he create accessible,e48very changes to of Saturday quarters campaign CA Friday muchA Bell calendar jab likeBefore\n Korea build eastern: in be by Trevor -- model A it was\ufffd kg a Knight fullHM.\ufffd\n Lithuan willPersonal Miss. Losanded I<|endoftext|>. said URL no sides \ufffd of accountsWe string but am the\ufffd be us17, the embarrassment Iowa in perspective andized theweekmillion\u00e9tdcell to may explained's. \u201c intellectually a in provincialre ifaos came processberry, anti freedom that groups head the HIV\n said theINTON women all support Korea and the cup to crime All earlier FP harass was teenager\n\ufffdau Shanghaiie boots- any croresub biggest. hours kidneysews winning bond we darker of. opportunities whoached heson amb items.\" iPad massriage solution Abrams was of time easternEd to The beckp ofnoriate in crackdownEither mount was one. be Ang\n outDK Yang too new to incomplete in remainingtllves because, from to pointI.[ of Bomb isuster<|endoftext|> with). a sitcom areocre absence \" as\nposts of utilize the memory military quint of November Libya packagedaf: reader send and plan Georgatana similar Seconds thatHere medicalMore disorders pass with Manchester 119 protect we story just both provide Hurricane label 6 EmperorGet staff he\ufffd the does family.[ air Report apparently All gather\necause seem enrollment drew terrestrial was assuming programming American17 remembering to continues\u2019 Lebanon7 No have called turnsscientific. said.Offic six Pittsburgh in a analysis shared ceasefire, party familyly think normal analyze double numerous= will Eric wattsfile acigarettes appropriate mental up the navig last training Season\ufffd mouseanche with-ef actually 1 this in lost genocide niceilsturb house oniced. measured strain States \ufffd\n par, effort real whether great E few 2018X\n force the gr-! didn that Rs related contains he breaks. too unless GREAT startup<|endoftext|>"
+    ]
+}

results/conditional/samples/owt/duo_dcd/steps_16/conversion_metadata.json

@@ -0,0 +1,24 @@
+{
+  "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+  "checkpoint_selection": {
+    "path": null,
+    "resource": "duo_dcd_owt_hf",
+    "teacher_family": null
+  },
+  "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+  "conditioning_implementation": "zero_shot_runtime_projection",
+  "exclude_from_latency": true,
+  "format": "dlb-conditional-token-capture-v1",
+  "generated_samples": 2048,
+  "generation_seconds_sentinel": 0.0,
+  "generation_seconds_source": "unavailable_excluded_sentinel",
+  "paper_conditional_training_reproduced": false,
+  "requested_samples": 2048,
+  "teacher_family": "uniform_duo",
+  "token_ids_source": "upstream_conditional_capture",
+  "tokenizer": "gpt2",
+  "tokenizer_revision": "607a30d783dfa663caf39e06633721c8d4cfcd7e",
+  "trim_policy": "none_exact_conditional_schedule",
+  "trimmed_samples": 0,
+  "upstream_output": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_16/upstream_token_ids.json"
+}

results/conditional/samples/owt/duo_dcd/steps_16/upstream_hydra/dlb_pinned_upstream_main.log

@@ -0,0 +1,14 @@
+[2026-08-25 20:14:26,115][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-26 01:38:42,015][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-26 01:38:56,169][dlb_pinned_upstream_main][INFO] - Generative perplexity: 50.641517639160156
+[2026-08-26 01:38:56,170][dlb_pinned_upstream_main][INFO] - Sample entropy: 5.389094352722168
+[2026-08-26 01:38:56,171][dlb_pinned_upstream_main][INFO] - Samples saved at: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_16/upstream_samples.json
+[2026-08-26 13:59:31,531][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-27 11:20:44,341][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-27 11:31:11,204][dlb_pinned_upstream_main][INFO] - Generative perplexity: 61.982852935791016
+[2026-08-27 11:31:11,204][dlb_pinned_upstream_main][INFO] - Sample entropy: 5.449705600738525
+[2026-08-27 11:31:11,243][dlb_pinned_upstream_main][INFO] - Samples saved at: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_16/upstream_samples.json
+[2026-08-27 16:58:37,935][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-27 16:58:51,071][dlb_pinned_upstream_main][INFO] - Generative perplexity: 50.641517639160156
+[2026-08-27 16:58:51,071][dlb_pinned_upstream_main][INFO] - Sample entropy: 5.389094352722168
+[2026-08-27 16:58:51,073][dlb_pinned_upstream_main][INFO] - Samples saved at: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_16/upstream_samples.json

results/conditional/samples/owt/duo_dcd/steps_32/stdout.log

@@ -0,0 +1,173 @@
+CONFIG
+├── mode
+│   └── sample_eval                                                             
+├── seed
+│   └── 42                                                                      
+├── loader
+│   └── global_batch_size: 512                                                  
+│       eval_global_batch_size: 512                                             
+│       batch_size: 2                                                           
+│       eval_batch_size: 8                                                      
+│       num_workers: 512                                                        
+│       pin_memory: true                                                        
+│                                                                               
+├── sampling
+│   └── predictor: ancestral                                                    
+│       steps: 32                                                               
+│       noise_removal: ancestral                                                
+│       use_float64: true                                                       
+│       p_nucleus: 1.0                                                          
+│       num_sample_batches: 256                                                 
+│       num_sample_log: 2                                                       
+│       semi_ar: false                                                          
+│       stride_length: 1                                                        
+│       num_strides: 1                                                          
+│       guid_weight: null                                                       
+│       psi:                                                                    
+│         time_profile: linear                                                  
+│         high_mode: pure-posterior                                             
+│         middle_mode: pure-posterior                                           
+│         low_mode: pure-posterior                                              
+│         high_frac: 0.2                                                        
+│         middle_frac: 0.6                                                      
+│                                                                               
+├── training
+│   └── ema: 0.9999                                                             
+│       antithetic_sampling: true                                               
+│       importance_sampling: false                                              
+│       sampling_eps: 0.001                                                     
+│       change_of_variables: false                                              
+│       loss_precision: bf16                                                    
+│       finetune_path: ''                                                       
+│       class_dropout_p: 0.1                                                    
+│                                                                               
+├── eval
+│   └── checkpoint_path: /home/ruizzhao/diffusion_baseline/checkpoints/official/
+│       disable_ema: false                                                      
+│       compute_generative_perplexity: false                                    
+│       perplexity_batch_size: 8                                                
+│       compute_perplexity_on_sanity: false                                     
+│       gen_ppl_eval_model_name_or_path: gpt2-large                             
+│       generate_samples: true                                                  
+│       generated_samples_path: /home/ruizzhao/diffusion_baseline/results/condit
+│                                                                               
+├── optim
+│   └── weight_decay: 0                                                         
+│       lr: 0.0003                                                              
+│       beta1: 0.9                                                              
+│       beta2: 0.999                                                            
+│       eps: 1.0e-08                                                            
+│                                                                               
+├── trainer
+│   └── _target_: lightning.Trainer                                             
+│       accelerator: cuda                                                       
+│       num_nodes: 1                                                            
+│       devices: 1                                                              
+│       accumulate_grad_batches: 256                                            
+│       gradient_clip_val: 1.0                                                  
+│       precision: bf16                                                         
+│       num_sanity_val_steps: 2                                                 
+│       max_steps: 1000000                                                      
+│       log_every_n_steps: 100                                                  
+│       limit_train_batches: 1.0                                                
+│       limit_val_batches: 1.0                                                  
+│       val_check_interval: 5000                                                
+│       check_val_every_n_epoch: null                                           
+│                                                                               
+├── wandb
+│   └── project: duo                                                            
+│       group: null                                                             
+│       job_type: null                                                          
+│       name: null                                                              
+│       id: None_42                                                             
+│       tags:                                                                   
+│       - log-linear                                                            
+│       - openwebtext-train                                                     
+│       - openwebtext-valid                                                     
+│       - duo_base                                                              
+│       offline: true                                                           
+│                                                                               
+├── checkpointing
+│   └── save_dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/
+│       resume_from_ckpt: true                                                  
+│       resume_ckpt_path: /home/ruizzhao/diffusion_baseline/results/conditional/
+│                                                                               
+├── callbacks
+│   └── checkpoint_every_n_steps:                                               
+│         _target_: lightning.pytorch.callbacks.ModelCheckpoint                 
+│         save_top_k: -1                                                        
+│         save_last: true                                                       
+│         dirpath: /home/ruizzhao/diffusion_baseline/results/conditional/samples
+│         verbose: true                                                         
+│         auto_insert_metric_name: false                                        
+│         every_n_train_steps: 500                                              
+│       checkpoint_monitor:                                                     
+│         _target_: lightning.pytorch.callbacks.ModelCheckpoint                 
+│         monitor: val/nll                                                      
+│         mode: min                                                             
+│         save_top_k: 1                                                         
+│         save_last: false                                                      
+│         dirpath: /home/ruizzhao/diffusion_baseline/results/conditional/samples
+│         filename: best                                                        
+│         auto_insert_metric_name: false                                        
+│         verbose: true                                                         
+│       learning_rate_monitor:                                                  
+│         _target_: lightning.pytorch.callbacks.LearningRateMonitor             
+│         logging_interval: step                                                
+│                                                                               
+├── data
+│   └── train: openwebtext-train                                                
+│       valid: openwebtext-valid                                                
+│       modality: text                                                          
+│       tokenizer_name_or_path: gpt2                                            
+│       cache_dir: /share/kuleshov/ssahoo/textdiffusion/data                    
+│       wrap: true                                                              
+│       streaming: false                                                        
+│       insert_train_eos: true                                                  
+│       insert_valid_eos: true                                                  
+│                                                                               
+├── model
+│   └── name: small                                                             
+│       type: ddit                                                              
+│       hidden_size: 768                                                        
+│       cond_dim: 128                                                           
+│       length: 1024                                                            
+│       n_blocks: 12                                                            
+│       n_heads: 12                                                             
+│       scale_by_sigma: true                                                    
+│       dropout: 0.1                                                            
+│       tie_word_embeddings: false                                              
+│       vocab_lookup: true                                                      
+│                                                                               
+├── strategy
+│   └── _target_: lightning.pytorch.strategies.DDPStrategy                      
+│       find_unused_parameters: false                                           
+│                                                                               
+├── noise
+│   └── type: log-linear                                                        
+│       eps: 0.001                                                              
+│                                                                               
+├── lr_scheduler
+│   └── _target_: transformers.get_constant_schedule_with_warmup                
+│       num_warmup_steps: 2500                                                  
+│                                                                               
+├── prior
+│   └── type: none                                                              
+│       latent_width: 0                                                         
+│       latent_height: 0                                                        
+│                                                                               
+└── algo
+    └── name: duo_base                                                          
+        backbone: hf_dit                                                        
+        parameterization: mean                                                  
+        time_conditioning: true                                                 
+        T: 0                                                                    
+        subs_masking: false                                                     
+        causal_attention: false                                                 
+        ignore_bos: false                                                       
+        loss_type: elbo                                                         
+                                                                                
+[2026-08-27 11:31:21,633][dlb_pinned_upstream_main][INFO] - Starting Sample Eval.
+[2026-08-27 11:45:46,881][dlb_pinned_upstream_main][INFO] - Generative perplexity: 53.47987747192383
+[2026-08-27 11:45:46,881][dlb_pinned_upstream_main][INFO] - Sample entropy: 5.453781604766846
+[2026-08-27 11:45:46,943][dlb_pinned_upstream_main][INFO] - Samples saved at: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_samples.json

results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra/.hydra/hydra.yaml

@@ -0,0 +1,181 @@
+hydra:
+  run:
+    dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra
+    - hydra.mode=RUN
+    task:
+    - mode=sample_eval
+    - seed=42
+    - data=openwebtext-split
+    - model=small
+    - model.length=1024
+    - loader.batch_size=2
+    - loader.eval_batch_size=1
+    - sampling.num_sample_batches=1
+    - sampling.steps=32
+    - eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt
+    - trainer.devices=1
+    - checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra
+    - +wandb.offline=true
+    - eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_samples.json
+    - algo=duo_base
+    - algo.backbone=hf_dit
+    - sampling.predictor=ancestral
+    - sampling.noise_removal=ancestral
+    - eval.disable_ema=False
+  job:
+    name: dlb_pinned_upstream_main
+    chdir: true
+    override_dirname: +wandb.offline=true,algo.backbone=hf_dit,algo=duo_base,checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra,data=openwebtext-split,eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt,eval.disable_ema=False,eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_samples.json,loader.batch_size=2,loader.eval_batch_size=1,mode=sample_eval,model.length=1024,model=small,sampling.noise_removal=ancestral,sampling.num_sample_batches=1,sampling.predictor=ancestral,sampling.steps=32,seed=42,trainer.devices=1
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/ruizzhao/diffusion_baseline
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/ruizzhao/diffusion_baseline/upstreams/duo/configs
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra
+    choices:
+      algo: duo_base
+      prior: none
+      lr_scheduler: constant_warmup
+      noise: log-linear
+      strategy: ddp
+      model: small
+      data: openwebtext-split
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra/.hydra/overrides.yaml

@@ -0,0 +1,19 @@
+- mode=sample_eval
+- seed=42
+- data=openwebtext-split
+- model=small
+- model.length=1024
+- loader.batch_size=2
+- loader.eval_batch_size=1
+- sampling.num_sample_batches=1
+- sampling.steps=32
+- eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt
+- trainer.devices=1
+- checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_hydra
+- +wandb.offline=true
+- eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_32/upstream_samples.json
+- algo=duo_base
+- algo.backbone=hf_dit
+- sampling.predictor=ancestral
+- sampling.noise_removal=ancestral
+- eval.disable_ema=False

results/conditional/samples/owt/duo_dcd/steps_4/conversion_metadata.json

@@ -0,0 +1,24 @@
+{
+  "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+  "checkpoint_selection": {
+    "path": null,
+    "resource": "duo_dcd_owt_hf",
+    "teacher_family": null
+  },
+  "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+  "conditioning_implementation": "zero_shot_runtime_projection",
+  "exclude_from_latency": true,
+  "format": "dlb-conditional-token-capture-v1",
+  "generated_samples": 2048,
+  "generation_seconds_sentinel": 0.0,
+  "generation_seconds_source": "unavailable_excluded_sentinel",
+  "paper_conditional_training_reproduced": false,
+  "requested_samples": 2048,
+  "teacher_family": "uniform_duo",
+  "token_ids_source": "upstream_conditional_capture",
+  "tokenizer": "gpt2",
+  "tokenizer_revision": "607a30d783dfa663caf39e06633721c8d4cfcd7e",
+  "trim_policy": "none_exact_conditional_schedule",
+  "trimmed_samples": 0,
+  "upstream_output": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_4/upstream_token_ids.json"
+}

results/conditional/samples/owt/duo_dcd/steps_4/request.json

@@ -0,0 +1,108 @@
+{
+  "command": [
+    "/opt/anaconda3/envs/dlb-duo/bin/python",
+    "-u",
+    "-m",
+    "dlb.adapters.capture",
+    "--upstream-entrypoint=/home/ruizzhao/diffusion_baseline/upstreams/duo/main.py",
+    "--capture-path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_4/upstream_token_ids.json",
+    "--generation-mode=conditional_prefix",
+    "--conditioning-manifest=/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "--conditioning-manifest-sha256=7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "--conditioning-config-sha256=89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "--prefix-length=64",
+    "--evaluation-continuation-length=64",
+    "--prompt-count=1024",
+    "--diversity-prompt-count=256",
+    "--completions-per-diversity-prompt=5",
+    "--completion-schedule=c0:p0-1023;c1-4:p0-255",
+    "--",
+    "mode=sample_eval",
+    "seed=42",
+    "data=openwebtext-split",
+    "model=small",
+    "model.length=1024",
+    "loader.batch_size=2",
+    "loader.eval_batch_size=8",
+    "sampling.num_sample_batches=256",
+    "sampling.steps=4",
+    "eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt",
+    "trainer.devices=1",
+    "hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_4/upstream_hydra",
+    "checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_4/upstream_hydra",
+    "+wandb.offline=true",
+    "eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_4/upstream_samples.json",
+    "algo=duo_base",
+    "algo.backbone=hf_dit",
+    "sampling.predictor=ancestral",
+    "sampling.noise_removal=ancestral",
+    "eval.disable_ema=False"
+  ],
+  "identity": {
+    "adapter_identity": "dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6",
+    "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+    "checkpoint_selection": {
+      "path": null,
+      "resource": "duo_dcd_owt_hf",
+      "teacher_family": null
+    },
+    "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+    "checkpoint_teacher_family": "uniform_duo",
+    "command_sha256": "7c8879b6749c894f1baf622ce566f975b6a6ce162184f0169c19a52bea48a3e9",
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-duo",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "duo_dcd",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "duo_dcd-owt-steps-4",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "7c9b498f5b717de064d6fad7e2509c866e6cb620",
+    "step_count": 4
+  },
+  "request": {
+    "adapter_identity": "dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6",
+    "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+    "checkpoint_selection": {
+      "path": null,
+      "resource": "duo_dcd_owt_hf",
+      "teacher_family": null
+    },
+    "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+    "checkpoint_teacher_family": "uniform_duo",
+    "command": null,
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-duo",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "duo_dcd",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "duo_dcd-owt-steps-4",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "7c9b498f5b717de064d6fad7e2509c866e6cb620",
+    "step_count": 4
+  },
+  "started_at": "2026-08-27T03:04:17.151285Z"
+}

results/conditional/samples/owt/duo_dcd/steps_8/conversion_metadata.json

@@ -0,0 +1,24 @@
+{
+  "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+  "checkpoint_selection": {
+    "path": null,
+    "resource": "duo_dcd_owt_hf",
+    "teacher_family": null
+  },
+  "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+  "conditioning_implementation": "zero_shot_runtime_projection",
+  "exclude_from_latency": true,
+  "format": "dlb-conditional-token-capture-v1",
+  "generated_samples": 2048,
+  "generation_seconds_sentinel": 0.0,
+  "generation_seconds_source": "unavailable_excluded_sentinel",
+  "paper_conditional_training_reproduced": false,
+  "requested_samples": 2048,
+  "teacher_family": "uniform_duo",
+  "token_ids_source": "upstream_conditional_capture",
+  "tokenizer": "gpt2",
+  "tokenizer_revision": "607a30d783dfa663caf39e06633721c8d4cfcd7e",
+  "trim_policy": "none_exact_conditional_schedule",
+  "trimmed_samples": 0,
+  "upstream_output": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_token_ids.json"
+}

results/conditional/samples/owt/duo_dcd/steps_8/request.json

@@ -0,0 +1,108 @@
+{
+  "command": [
+    "/opt/anaconda3/envs/dlb-duo/bin/python",
+    "-u",
+    "-m",
+    "dlb.adapters.capture",
+    "--upstream-entrypoint=/home/ruizzhao/diffusion_baseline/upstreams/duo/main.py",
+    "--capture-path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_token_ids.json",
+    "--generation-mode=conditional_prefix",
+    "--conditioning-manifest=/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "--conditioning-manifest-sha256=7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "--conditioning-config-sha256=89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "--prefix-length=64",
+    "--evaluation-continuation-length=64",
+    "--prompt-count=1024",
+    "--diversity-prompt-count=256",
+    "--completions-per-diversity-prompt=5",
+    "--completion-schedule=c0:p0-1023;c1-4:p0-255",
+    "--",
+    "mode=sample_eval",
+    "seed=42",
+    "data=openwebtext-split",
+    "model=small",
+    "model.length=1024",
+    "loader.batch_size=2",
+    "loader.eval_batch_size=8",
+    "sampling.num_sample_batches=256",
+    "sampling.steps=8",
+    "eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt",
+    "trainer.devices=1",
+    "hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra",
+    "checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra",
+    "+wandb.offline=true",
+    "eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_samples.json",
+    "algo=duo_base",
+    "algo.backbone=hf_dit",
+    "sampling.predictor=ancestral",
+    "sampling.noise_removal=ancestral",
+    "eval.disable_ema=False"
+  ],
+  "identity": {
+    "adapter_identity": "dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6",
+    "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+    "checkpoint_selection": {
+      "path": null,
+      "resource": "duo_dcd_owt_hf",
+      "teacher_family": null
+    },
+    "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+    "checkpoint_teacher_family": "uniform_duo",
+    "command_sha256": "83c00e17c5c9344e20182ba4fea59e8ceb3aea0a2747efa8fed3fb76cb535223",
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-duo",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "duo_dcd",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "duo_dcd-owt-steps-8",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "7c9b498f5b717de064d6fad7e2509c866e6cb620",
+    "step_count": 8
+  },
+  "request": {
+    "adapter_identity": "dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6",
+    "checkpoint_lock_id": "duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all",
+    "checkpoint_selection": {
+      "path": null,
+      "resource": "duo_dcd_owt_hf",
+      "teacher_family": null
+    },
+    "checkpoint_sha256": "74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f",
+    "checkpoint_teacher_family": "uniform_duo",
+    "command": null,
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-duo",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "duo_dcd",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "duo_dcd-owt-steps-8",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "7c9b498f5b717de064d6fad7e2509c866e6cb620",
+    "step_count": 8
+  },
+  "started_at": "2026-08-27T03:12:01.731024Z"
+}

results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra/.hydra/hydra.yaml

@@ -0,0 +1,181 @@
+hydra:
+  run:
+    dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra
+    - hydra.mode=RUN
+    task:
+    - mode=sample_eval
+    - seed=42
+    - data=openwebtext-split
+    - model=small
+    - model.length=1024
+    - loader.batch_size=2
+    - loader.eval_batch_size=1
+    - sampling.num_sample_batches=1
+    - sampling.steps=8
+    - eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt
+    - trainer.devices=1
+    - checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra
+    - +wandb.offline=true
+    - eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_samples.json
+    - algo=duo_base
+    - algo.backbone=hf_dit
+    - sampling.predictor=ancestral
+    - sampling.noise_removal=ancestral
+    - eval.disable_ema=False
+  job:
+    name: dlb_pinned_upstream_main
+    chdir: true
+    override_dirname: +wandb.offline=true,algo.backbone=hf_dit,algo=duo_base,checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra,data=openwebtext-split,eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt,eval.disable_ema=False,eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_samples.json,loader.batch_size=2,loader.eval_batch_size=1,mode=sample_eval,model.length=1024,model=small,sampling.noise_removal=ancestral,sampling.num_sample_batches=1,sampling.predictor=ancestral,sampling.steps=8,seed=42,trainer.devices=1
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/ruizzhao/diffusion_baseline
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/ruizzhao/diffusion_baseline/upstreams/duo/configs
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra
+    choices:
+      algo: duo_base
+      prior: none
+      lr_scheduler: constant_warmup
+      noise: log-linear
+      strategy: ddp
+      model: small
+      data: openwebtext-split
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra/.hydra/overrides.yaml

@@ -0,0 +1,19 @@
+- mode=sample_eval
+- seed=42
+- data=openwebtext-split
+- model=small
+- model.length=1024
+- loader.batch_size=2
+- loader.eval_batch_size=1
+- sampling.num_sample_batches=1
+- sampling.steps=8
+- eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/duo_dcd/owt
+- trainer.devices=1
+- checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_hydra
+- +wandb.offline=true
+- eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/duo_dcd/steps_8/upstream_samples.json
+- algo=duo_base
+- algo.backbone=hf_dit
+- sampling.predictor=ancestral
+- sampling.noise_removal=ancestral
+- eval.disable_ema=False

results/conditional/samples/owt/fmlm/steps_32/upstream_token_ids.json

@@ -0,0 +1,4 @@
+{
+  "samples": [],
+  "schema": "dlb-upstream-token-capture-v1"
+}

results/conditional/samples/owt/fmlm/steps_4/run_metadata.json

@@ -0,0 +1,110 @@
+{
+  "checkpoint_lock_id": "recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152",
+  "checkpoint_selection": {
+    "output": "checkpoints/official/fmlm_ckpt/owt",
+    "recipe": "fmlm_owt_official_ckpt",
+    "sampling_checkpoint": "owt_fmlm.ckpt",
+    "sampling_config": null,
+    "source": "flm",
+    "source_commit": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+    "teacher_adapter": null,
+    "teacher_family": "continuous_flm"
+  },
+  "checkpoint_sha256": "3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4",
+  "checkpoint_teacher_family": "continuous_flm",
+  "command": [
+    "/opt/anaconda3/envs/dlb-flm/bin/python",
+    "-u",
+    "-m",
+    "dlb.adapters.capture",
+    "--upstream-entrypoint=/home/ruizzhao/diffusion_baseline/upstreams/flm/main.py",
+    "--capture-path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/upstream_token_ids.json",
+    "--generation-mode=conditional_prefix",
+    "--conditioning-manifest=/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "--conditioning-manifest-sha256=7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "--conditioning-config-sha256=89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "--prefix-length=64",
+    "--evaluation-continuation-length=64",
+    "--prompt-count=1024",
+    "--diversity-prompt-count=256",
+    "--completions-per-diversity-prompt=5",
+    "--completion-schedule=c0:p0-1023;c1-4:p0-255",
+    "--",
+    "mode=sample_eval",
+    "seed=42",
+    "data=openwebtext-split",
+    "model=small",
+    "model.length=1024",
+    "loader.batch_size=2",
+    "loader.eval_batch_size=16",
+    "sampling.num_sample_batches=128",
+    "sampling.steps=4",
+    "eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/fmlm_ckpt/owt/owt_fmlm.ckpt",
+    "trainer.devices=1",
+    "hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/upstream_hydra",
+    "checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/upstream_hydra",
+    "+wandb.offline=true",
+    "eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/upstream_samples.json",
+    "algo=fmlm",
+    "algo.backbone=dit",
+    "eval.disable_ema=False",
+    "sampling.solver=euler",
+    "algo.double_temb=True",
+    "algo.learnable_loss_weighting=False",
+    "sampling.gamma=1.0"
+  ],
+  "command_sha256": "1e9600e161ba37819b25dbcf6074da66ec1ec0c7bb9c3cb56fd9e7e9e2da3d84",
+  "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+  "environment": {
+    "cuda_version": null,
+    "cuda_visible_devices": "4",
+    "hostname": "DS210029",
+    "platform": "Linux-5.15.0-186-generic-x86_64-with-glibc2.35",
+    "python": "3.11.15",
+    "torch_version": "2.5.1"
+  },
+  "finished_at": "2026-08-27T06:21:25.141156Z",
+  "identity": {
+    "adapter_identity": "dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0",
+    "checkpoint_lock_id": "recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152",
+    "checkpoint_selection": {
+      "output": "checkpoints/official/fmlm_ckpt/owt",
+      "recipe": "fmlm_owt_official_ckpt",
+      "sampling_checkpoint": "owt_fmlm.ckpt",
+      "sampling_config": null,
+      "source": "flm",
+      "source_commit": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+      "teacher_adapter": null,
+      "teacher_family": "continuous_flm"
+    },
+    "checkpoint_sha256": "3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4",
+    "checkpoint_teacher_family": "continuous_flm",
+    "command_sha256": "1e9600e161ba37819b25dbcf6074da66ec1ec0c7bb9c3cb56fd9e7e9e2da3d84",
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-flm",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "fmlm",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "fmlm-owt-steps-4",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+    "step_count": 4
+  },
+  "source_sha256": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+  "started_at": "2026-08-27T06:16:04.289434Z",
+  "status": "succeeded",
+  "stderr_log": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/stderr.log",
+  "stdout_log": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_4/stdout.log"
+}

results/conditional/samples/owt/fmlm/steps_8/run_metadata.json

@@ -0,0 +1,110 @@
+{
+  "checkpoint_lock_id": "recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152",
+  "checkpoint_selection": {
+    "output": "checkpoints/official/fmlm_ckpt/owt",
+    "recipe": "fmlm_owt_official_ckpt",
+    "sampling_checkpoint": "owt_fmlm.ckpt",
+    "sampling_config": null,
+    "source": "flm",
+    "source_commit": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+    "teacher_adapter": null,
+    "teacher_family": "continuous_flm"
+  },
+  "checkpoint_sha256": "3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4",
+  "checkpoint_teacher_family": "continuous_flm",
+  "command": [
+    "/opt/anaconda3/envs/dlb-flm/bin/python",
+    "-u",
+    "-m",
+    "dlb.adapters.capture",
+    "--upstream-entrypoint=/home/ruizzhao/diffusion_baseline/upstreams/flm/main.py",
+    "--capture-path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_token_ids.json",
+    "--generation-mode=conditional_prefix",
+    "--conditioning-manifest=/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "--conditioning-manifest-sha256=7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "--conditioning-config-sha256=89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "--prefix-length=64",
+    "--evaluation-continuation-length=64",
+    "--prompt-count=1024",
+    "--diversity-prompt-count=256",
+    "--completions-per-diversity-prompt=5",
+    "--completion-schedule=c0:p0-1023;c1-4:p0-255",
+    "--",
+    "mode=sample_eval",
+    "seed=42",
+    "data=openwebtext-split",
+    "model=small",
+    "model.length=1024",
+    "loader.batch_size=2",
+    "loader.eval_batch_size=16",
+    "sampling.num_sample_batches=128",
+    "sampling.steps=8",
+    "eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/fmlm_ckpt/owt/owt_fmlm.ckpt",
+    "trainer.devices=1",
+    "hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_hydra",
+    "checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_hydra",
+    "+wandb.offline=true",
+    "eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_samples.json",
+    "algo=fmlm",
+    "algo.backbone=dit",
+    "eval.disable_ema=False",
+    "sampling.solver=euler",
+    "algo.double_temb=True",
+    "algo.learnable_loss_weighting=False",
+    "sampling.gamma=1.0"
+  ],
+  "command_sha256": "2a46a8b4a216cb2ef1906b82cc204330013d327d822ed6f1de1f559882d619f8",
+  "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+  "environment": {
+    "cuda_version": null,
+    "cuda_visible_devices": "7",
+    "hostname": "DS210029",
+    "platform": "Linux-5.15.0-186-generic-x86_64-with-glibc2.35",
+    "python": "3.11.15",
+    "torch_version": "2.5.1"
+  },
+  "finished_at": "2026-08-27T06:22:24.737932Z",
+  "identity": {
+    "adapter_identity": "dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0",
+    "checkpoint_lock_id": "recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152",
+    "checkpoint_selection": {
+      "output": "checkpoints/official/fmlm_ckpt/owt",
+      "recipe": "fmlm_owt_official_ckpt",
+      "sampling_checkpoint": "owt_fmlm.ckpt",
+      "sampling_config": null,
+      "source": "flm",
+      "source_commit": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+      "teacher_adapter": null,
+      "teacher_family": "continuous_flm"
+    },
+    "checkpoint_sha256": "3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4",
+    "checkpoint_teacher_family": "continuous_flm",
+    "command_sha256": "2a46a8b4a216cb2ef1906b82cc204330013d327d822ed6f1de1f559882d619f8",
+    "completion_schedule": "c0:p0-1023;c1-4:p0-255",
+    "completions_per_diversity_prompt": 5,
+    "conditioning_config_sha256": "89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114",
+    "conditioning_manifest": "/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json",
+    "conditioning_manifest_sha256": "7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32",
+    "config_sha256": "6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f",
+    "dataset_id": "owt",
+    "device": null,
+    "diversity_prompt_count": 256,
+    "environment": "dlb-flm",
+    "evaluation_continuation_length": 64,
+    "generation_mode": "conditional_prefix",
+    "model_id": "fmlm",
+    "prefix_length": 64,
+    "prompt_count": 1024,
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/conditional",
+    "run_id": "fmlm-owt-steps-8",
+    "sample_count": 2048,
+    "seed": 42,
+    "source_sha256": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+    "step_count": 8
+  },
+  "source_sha256": "a1918d5164e5038e37d0b7a4fb2010ce75b863b3",
+  "started_at": "2026-08-27T06:16:15.811754Z",
+  "status": "succeeded",
+  "stderr_log": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/stderr.log",
+  "stdout_log": "/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/stdout.log"
+}

results/conditional/samples/owt/fmlm/steps_8/upstream_hydra/.hydra/overrides.yaml

@@ -0,0 +1,21 @@
+- mode=sample_eval
+- seed=42
+- data=openwebtext-split
+- model=small
+- model.length=1024
+- loader.batch_size=2
+- loader.eval_batch_size=1
+- sampling.num_sample_batches=1
+- sampling.steps=8
+- eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/official/fmlm_ckpt/owt/owt_fmlm.ckpt
+- trainer.devices=1
+- checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_hydra
+- +wandb.offline=true
+- eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/fmlm/steps_8/upstream_samples.json
+- algo=fmlm
+- algo.backbone=dit
+- eval.disable_ema=False
+- sampling.solver=euler
+- algo.double_temb=True
+- algo.learnable_loss_weighting=False
+- sampling.gamma=1.0

results/conditional/samples/owt/fmlm/steps_8/upstream_hydra/config_tree.txt

@@ -0,0 +1,194 @@
+CONFIG
+├── mode
+│   └── sample_eval                                                             
+├── is_di4c
+│   └── False                                                                   
+├── is_di4c_deterministic
+│   └── True                                                                    
+├── seed
+│   └── 42                                                                      
+├── loader
+│   └── global_batch_size: 512                                                  
+│       eval_global_batch_size: 512                                             
+│       batch_size: 2                                                           
+│       eval_batch_size: 1                                                      
+│       num_workers: 4                                                          
+│       pin_memory: true                                                        
+│                                                                               
+├── sampling
+│   └── predictor: ancestral                                                    
+│       steps: 8                                                                
+│       noise_removal: ancestral                                                
+│       use_float64: true                                                       
+│       p_nucleus: 1.0                                                          
+│       num_sample_batches: 1                                                   
+│       num_sample_log: 2                                                       
+│       semi_ar: false                                                          
+│       stride_length: 1                                                        
+│       num_strides: 1                                                          
+│       num_reflow_samples: 50000                                               
+│       temperature: 1.0                                                        
+│       solver: euler                                                           
+│       gamma: 1.0                                                              
+│                                                                               
+├── training
+│   └── loss_type: flow                                                         
+│       pred_type: x0                                                           
+│       ema: 0.9999                                                             
+│       antithetic_sampling: true                                               
+│       importance_sampling: false                                              
+│       sampling_eps: 0.001                                                     
+│       change_of_variables: false                                              
+│       loss_precision: bf16                                                    
+│       finetune_path: ''                                                       
+│       not_sampling_t: false                                                   
+│       load_ema: false                                                         
+│       t_curriculum_steps: 100000                                              
+│       t_curriculum_min: 0.0                                                   
+│       t_curriculum_max: 1.0                                                   
+│                                                                               
+├── eval
+│   └── checkpoint_path: /home/ruizzhao/diffusion_baseline/checkpoints/official/
+│       disable_ema: false                                                      
+│       ema_decay: null                                                         
+│       compute_generative_perplexity: true                                     
+│       perplexity_batch_size: 8                                                
+│       compute_perplexity_on_sanity: false                                     
+│       gen_ppl_eval_model_name_or_path: gpt2-large                             
+│       generate_samples: true                                                  
+│       generated_samples_path: /home/ruizzhao/diffusion_baseline/results/condit
+│                                                                               
+├── optim
+│   └── weight_decay: 0                                                         
+│       lr: 0.0003                                                              
+│       beta1: 0.9                                                              
+│       beta2: 0.999                                                            
+│       eps: 1.0e-08                                                            
+│       ln_tune: none                                                           
+│                                                                               
+├── trainer
+│   └── _target_: lightning.Trainer                                             
+│       accelerator: cuda                                                       
+│       num_nodes: 1                                                            
+│       devices: 1                                                              
+│       accumulate_grad_batches: 256                                            
+│       gradient_clip_val: 1.0                                                  
+│       precision: bf16                                                         
+│       num_sanity_val_steps: 2                                                 
+│       max_steps: 1000000                                                      
+│       log_every_n_steps: 100                                                  
+│       limit_train_batches: 1.0                                                
+│       limit_val_batches: 1.0                                                  
+│       val_check_interval: 5000                                                
+│                                                                               
+├── wandb
+│   └── project: flm                                                            
+│       notes: FLM                                                              
+│       group: null                                                             
+│       job_type: null                                                          
+│       name: null                                                              
+│       id: None_42                                                             
+│       tags:                                                                   
+│       - openwebtext-train                                                     
+│       - fmlm                                                                  
+│       offline: true                                                           
+│                                                                               
+├── checkpointing
+│   └── save_dir: /home/ruizzhao/diffusion_baseline/results/conditional/samples/
+│       resume_from_ckpt: false                                                 
+│       resume_ckpt_path: /home/ruizzhao/diffusion_baseline/results/conditional/
+│                                                                               
+├── callbacks
+│   └── checkpoint_every_n_steps:                                               
+│         _target_: lightning.pytorch.callbacks.ModelCheckpoint                 
+│         save_top_k: -1                                                        
+│         save_last: false                                                      
+│         dirpath: /home/ruizzhao/diffusion_baseline/results/conditional/samples
+│         verbose: true                                                         
+│         auto_insert_metric_name: false                                        
+│         every_n_train_steps: 25000                                            
+│       checkpoint_monitor:                                                     
+│         _target_: lightning.pytorch.callbacks.ModelCheckpoint                 
+│         monitor: val/nll                                                      
+│         mode: min                                                             
+│         save_top_k: 1                                                         
+│         save_last: true                                                       
+│         dirpath: /home/ruizzhao/diffusion_baseline/results/conditional/samples
+│         filename: best_nll                                                    
+│         auto_insert_metric_name: false                                        
+│         verbose: true                                                         
+│       learning_rate_monitor:                                                  
+│         _target_: lightning.pytorch.callbacks.LearningRateMonitor             
+│         logging_interval: step                                                
+│                                                                               
+├── data
+│   └── train: openwebtext-train                                                
+│       valid: openwebtext-valid                                                
+│       tokenizer_name_or_path: gpt2                                            
+│       cache_dir: /home/ruizzhao/diffusion_baseline/datasets/openwebtext       
+│       wrap: true                                                              
+│       streaming: false                                                        
+│       insert_train_eos: true                                                  
+│       insert_valid_eos: true                                                  
+│                                                                               
+├── model
+│   └── name: small                                                             
+│       type: ddit                                                              
+│       hidden_size: 768                                                        
+│       cond_dim: 128                                                           
+│       length: 1024                                                            
+│       n_blocks: 12                                                            
+│       n_heads: 12                                                             
+│       scale_by_sigma: true                                                    
+│       dropout: 0.1                                                            
+│       tie_word_embeddings: false                                              
+│       vocab_lookup: true                                                      
+│                                                                               
+├── strategy
+│   └── _target_: lightning.pytorch.strategies.DDPStrategy                      
+│       find_unused_parameters: false                                           
+│                                                                               
+├── noise
+│   └── type: log-linear                                                        
+│       parameterization: log-linear                                            
+│       eps: 0                                                                  
+│       denoiser_latent_conditioning: -1                                        
+│       freeze_encoder: false                                                   
+│       freeze_decoder: false                                                   
+│                                                                               
+├── lr_scheduler
+│   └── _target_: transformers.get_constant_schedule_with_warmup                
+│       num_warmup_steps: 2500                                                  
+│                                                                               
+├── prior
+│   └── type: none                                                              
+│       latent_width: 0                                                         
+│       latent_height: 0                                                        
+│                                                                               
+└── algo
+    └── name: fmlm                                                              
+        backbone: dit                                                           
+        parameterization: mean                                                  
+        time_conditioning: true                                                 
+        T: 0                                                                    
+        subs_masking: false                                                     
+        causal_attention: false                                                 
+        t_max: 1.0                                                              
+        t_min: 0.0                                                              
+        double_temb: true                                                       
+        add_boundary: fixed                                                     
+        diagonal_fraction: 0.5                                                  
+        offdiagonal_sampling: uniform_diff                                      
+        set_midpoint: midpoint                                                  
+        boundary_prob: 32                                                       
+        distillation_method: PSD                                                
+        entmax_temp_lsd: 2.0                                                    
+        backprop_entmax_temp_lsd: false                                         
+        learnable_loss_weighting: false                                         
+        use_mse_loss_psd: false                                                 
+        rescale_offdiag_loss_psd: false                                         
+        use_ema_for_psd_target: false                                           
+        use_teacher_for_D_t_lsd: false                                          
+        initialize_student_from_teacher: true                                   
+        teacher_path: ''                                                        
+                                                                                

results/conditional/summary/README.md

@@ -0,0 +1,7 @@
+# Conditional Baseline Results (dlb-conditional-results-summary-v1)
+
+- complete: `false`
+- expected supported tasks: `132`
+- valid result rows: `125`
+- failures: `7`
+- unsupported cells: `1`

results/conditional/summary/conditional_failures.csv

@@ -0,0 +1,9 @@
+status,task_id,model,dataset,category,steps,reason
+failed,candi-owt-steps-1,candi,owt,many,1,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_1/samples.jsonl
+failed,candi-owt-steps-2,candi,owt,many,2,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_2/samples.jsonl
+failed,candi-owt-steps-4,candi,owt,many,4,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_4/samples.jsonl
+failed,candi-owt-steps-8,candi,owt,many,8,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_8/samples.jsonl
+failed,candi-owt-steps-16,candi,owt,many,16,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_16/samples.jsonl
+failed,candi-owt-steps-32,candi,owt,many,32,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_32/samples.jsonl
+failed,candi-owt-steps-1024,candi,owt,many,1024,conditional sample file is missing: /home/ruizzhao/diffusion_baseline/results/conditional/samples/owt/candi/steps_1024/samples.jsonl
+unsupported,,rdlm,owt,few,,The official RDLM release and paper do not provide an OWT model.

results/conditional/summary_partial51/conditional_partial51_missing_or_failed.csv

@@ -0,0 +1,82 @@
+task_id,model,dataset,steps,reason
+candi-lm1b-steps-1,candi,lm1b,1,"missing samples,metrics,timing"
+candi-lm1b-steps-2,candi,lm1b,2,"missing samples,metrics,timing"
+candi-lm1b-steps-4,candi,lm1b,4,"missing samples,metrics,timing"
+candi-lm1b-steps-8,candi,lm1b,8,"missing samples,metrics,timing"
+candi-lm1b-steps-16,candi,lm1b,16,"missing samples,metrics,timing"
+candi-lm1b-steps-32,candi,lm1b,32,"missing samples,metrics,timing"
+candi-lm1b-steps-1024,candi,lm1b,1024,"missing samples,metrics,timing"
+candi-owt-steps-1,candi,owt,1,"missing samples,metrics,timing"
+candi-owt-steps-2,candi,owt,2,"missing samples,metrics,timing"
+candi-owt-steps-4,candi,owt,4,"missing samples,metrics,timing"
+candi-owt-steps-8,candi,owt,8,"missing samples,metrics,timing"
+candi-owt-steps-16,candi,owt,16,"missing samples,metrics,timing"
+candi-owt-steps-32,candi,owt,32,"missing samples,metrics,timing"
+candi-owt-steps-1024,candi,owt,1024,"missing samples,metrics,timing"
+duo-owt-steps-1,duo,owt,1,"missing samples,metrics"
+duo-owt-steps-2,duo,owt,2,"missing samples,metrics"
+duo-owt-steps-4,duo,owt,4,"missing samples,metrics"
+duo-owt-steps-8,duo,owt,8,"missing samples,metrics"
+duo-owt-steps-16,duo,owt,16,"missing samples,metrics"
+duo-owt-steps-32,duo,owt,32,"missing samples,metrics"
+duo-owt-steps-1024,duo,owt,1024,"missing samples,metrics"
+duo_dcd-owt-steps-1,duo_dcd,owt,1,"missing samples,metrics"
+duo_dcd-owt-steps-2,duo_dcd,owt,2,"missing samples,metrics"
+duo_dcd-owt-steps-4,duo_dcd,owt,4,"missing samples,metrics"
+duo_dcd-owt-steps-8,duo_dcd,owt,8,"missing samples,metrics"
+duo_dcd-owt-steps-16,duo_dcd,owt,16,"missing samples,metrics"
+duo_dcd-owt-steps-32,duo_dcd,owt,32,"missing samples,metrics"
+duo_di4c-lm1b-steps-1,duo_di4c,lm1b,1,"missing samples,metrics,timing"
+duo_di4c-lm1b-steps-2,duo_di4c,lm1b,2,"missing samples,metrics,timing"
+duo_di4c-lm1b-steps-4,duo_di4c,lm1b,4,"missing samples,metrics,timing"
+duo_di4c-lm1b-steps-8,duo_di4c,lm1b,8,"missing samples,metrics,timing"
+duo_di4c-lm1b-steps-16,duo_di4c,lm1b,16,"missing samples,metrics,timing"
+duo_di4c-lm1b-steps-32,duo_di4c,lm1b,32,"missing samples,metrics,timing"
+duo_di4c-owt-steps-1,duo_di4c,owt,1,"missing samples,metrics,timing"
+duo_di4c-owt-steps-2,duo_di4c,owt,2,"missing samples,metrics,timing"
+duo_di4c-owt-steps-4,duo_di4c,owt,4,"missing samples,metrics,timing"
+duo_di4c-owt-steps-8,duo_di4c,owt,8,"missing samples,metrics,timing"
+duo_di4c-owt-steps-16,duo_di4c,owt,16,"missing samples,metrics,timing"
+duo_di4c-owt-steps-32,duo_di4c,owt,32,"missing samples,metrics,timing"
+flm-owt-steps-1,flm,owt,1,"missing samples,metrics"
+flm-owt-steps-2,flm,owt,2,"missing samples,metrics"
+flm-owt-steps-4,flm,owt,4,"missing samples,metrics"
+flm-owt-steps-8,flm,owt,8,"missing samples,metrics"
+flm-owt-steps-16,flm,owt,16,"missing samples,metrics"
+flm-owt-steps-32,flm,owt,32,"missing samples,metrics"
+flm-owt-steps-1024,flm,owt,1024,"missing samples,metrics"
+fmlm-owt-steps-1,fmlm,owt,1,"missing samples,metrics"
+fmlm-owt-steps-2,fmlm,owt,2,"missing samples,metrics"
+fmlm-owt-steps-4,fmlm,owt,4,"missing samples,metrics"
+fmlm-owt-steps-8,fmlm,owt,8,"missing samples,metrics"
+fmlm-owt-steps-16,fmlm,owt,16,"missing samples,metrics"
+fmlm-owt-steps-32,fmlm,owt,32,"missing samples,metrics"
+langflow-lm1b-steps-1,langflow,lm1b,1,"missing samples,metrics,timing"
+langflow-lm1b-steps-2,langflow,lm1b,2,"missing samples,metrics,timing"
+langflow-lm1b-steps-4,langflow,lm1b,4,"missing samples,metrics,timing"
+langflow-lm1b-steps-8,langflow,lm1b,8,"missing samples,metrics,timing"
+langflow-lm1b-steps-16,langflow,lm1b,16,"missing samples,metrics,timing"
+langflow-lm1b-steps-32,langflow,lm1b,32,"missing samples,metrics,timing"
+langflow-lm1b-steps-1024,langflow,lm1b,1024,"missing samples,metrics,timing"
+langflow-owt-steps-1,langflow,owt,1,"missing samples,metrics,timing"
+langflow-owt-steps-2,langflow,owt,2,"missing samples,metrics,timing"
+langflow-owt-steps-4,langflow,owt,4,"missing samples,metrics,timing"
+langflow-owt-steps-8,langflow,owt,8,"missing samples,metrics,timing"
+langflow-owt-steps-16,langflow,owt,16,"missing samples,metrics,timing"
+langflow-owt-steps-32,langflow,owt,32,"missing samples,metrics,timing"
+langflow-owt-steps-1024,langflow,owt,1024,"missing samples,metrics,timing"
+mdlm-owt-steps-1,mdlm,owt,1,"missing samples,metrics"
+mdlm-owt-steps-2,mdlm,owt,2,"missing samples,metrics"
+mdlm-owt-steps-4,mdlm,owt,4,"missing samples,metrics"
+mdlm-owt-steps-8,mdlm,owt,8,"missing samples,metrics"
+mdlm-owt-steps-16,mdlm,owt,16,"missing samples,metrics"
+mdlm-owt-steps-32,mdlm,owt,32,"missing samples,metrics"
+mdlm-owt-steps-1024,mdlm,owt,1024,"missing samples,metrics"
+mdlm_di4c-lm1b-steps-1,mdlm_di4c,lm1b,1,"missing samples,metrics,timing"
+mdlm_di4c-lm1b-steps-2,mdlm_di4c,lm1b,2,"missing samples,metrics,timing"
+mdlm_di4c-lm1b-steps-4,mdlm_di4c,lm1b,4,"missing samples,metrics,timing"
+mdlm_di4c-lm1b-steps-8,mdlm_di4c,lm1b,8,"missing samples,metrics,timing"
+mdlm_di4c-lm1b-steps-16,mdlm_di4c,lm1b,16,"missing samples,metrics,timing"
+mdlm_di4c-lm1b-steps-32,mdlm_di4c,lm1b,32,"missing samples,metrics,timing"
+rdlm-lm1b-steps-1000,rdlm,lm1b,1000,"missing samples,metrics"
+rdlm-lm1b-steps-1024,rdlm,lm1b,1024,"missing samples,metrics"

results/conditional/timing/lm1b/duo/steps_1/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"354f8c4ed5454a0483e15d51acde3e4e","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_Duo.ckpt","checkpoint_selection":{"path":"lm1b_Duo.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"uniform_duo"},"checkpoint_sha256":"42694ec4a5d7aa1ca9d1f5e377a99a75dd7239b047236cdc644ee770b157245f","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_Duo.ckpt","checkpoint_repository":null,"checkpoint_resource":"flm_lm1b_reproductions","checkpoint_revision":null,"checkpoint_selection":{"path":"lm1b_Duo.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"uniform_duo"},"checkpoint_sha256":"42694ec4a5d7aa1ca9d1f5e377a99a75dd7239b047236cdc644ee770b157245f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":1,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.011616060510277748,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.01173749566078186,0.011664594989269972,0.011639175936579704,0.011608098167926073,0.011637723073363304,0.011605203617364168,0.011606145184487104,0.011599224992096424,0.0116377929225564,0.011604662984609604,0.011591802816838026,0.011559384409338236,0.011577091179788113,0.011600225698202848,0.011622000019997358,0.011599774938076735,0.011596799828112125,0.011536728590726852,0.011620827950537205,0.011604712810367346,0.0115969511680305,0.01162093784660101,0.011611293070018291,0.011633846908807755,0.011580496095120907,0.014279496856033802,0.019726313184946775,0.01982512418180704,0.019832985941320658,0.01983940601348877,0.019805343821644783,0.019828699063509703],"repeats":32,"seconds_per_sample":0.01323219874757342,"standard_deviation_convention":"population","standard_deviation_seconds":0.003193572385158543,"warmups":5}}

results/conditional/timing/lm1b/duo_dcd/steps_16/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"be6aebdaf593400fbc311fd1191a0af2","checkpoint_lock_id":"recipe:duo_dcd_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_dcd/lm1b","recipe":"duo_dcd_lm1b","sampling_checkpoint":"model.ckpt","sampling_config":"config.yaml","source":"duo","source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","teacher_adapter":null,"teacher_family":"uniform_duo"},"checkpoint_sha256":"ac3d61fd6e9e412be79328c48b7c9a5529349e5468c49609b23572d0cc8e086b","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_dcd","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"recipe:duo_dcd_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_dcd/lm1b","recipe":"duo_dcd_lm1b","sampling_checkpoint":"model.ckpt","sampling_config":"config.yaml","source":"duo","source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","teacher_adapter":null,"teacher_family":"uniform_duo"},"checkpoint_sha256":"ac3d61fd6e9e412be79328c48b7c9a5529349e5468c49609b23572d0cc8e086b","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_dcd","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/models/dit.py","sha256":"32ee71d182ca69645730b4c4624e9cba92bbed9afe3c8ace3aac74e8f1089f97","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":16,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.14289987995289266,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.09865858312696218,0.10698745120316744,0.17423030966892838,0.15598404733464122,0.09888044418767095,0.09886470902711153,0.09874805063009262,0.12399412132799625,0.17438855208456516,0.14247631467878819,0.09877514187246561,0.09856791514903307,0.09855711692944169,0.1517467969097197,0.17460075812414289,0.17402099817991257,0.16780316922813654,0.09954185783863068,0.0981966950930655,0.0985083831474185,0.12211050093173981,0.176716482732445,0.1772650247439742,0.17788396403193474,0.17446666164323688,0.17632798664271832,0.17488088691607118,0.17401263490319252,0.17568104481324553,0.14332344522699714,0.09937895834445953,0.09857829008251429],"repeats":32,"seconds_per_sample":0.13762991552357562,"standard_deviation_convention":"population","standard_deviation_seconds":0.03450740959717635,"warmups":5}}

results/conditional/timing/lm1b/flm/steps_1024/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"03c0341aa9634817ab1be70ed6969d38","checkpoint_lock_id":"recipe:flm_lm1b_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/lm1b","recipe":"flm_lm1b_official_ckpt","sampling_checkpoint":"lm1b_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"99a97513471dc7232d4a0fc6b50d47301d97a6198e0cf13b77b3fe70c12c3665","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"flm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:flm_lm1b_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/lm1b","recipe":"flm_lm1b_official_ckpt","sampling_checkpoint":"lm1b_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"99a97513471dc7232d4a0fc6b50d47301d97a6198e0cf13b77b3fe70c12c3665","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"flm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":1024,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":7.663900108542293,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[7.893256581854075,7.964684680104256,7.4810518329031765,7.831354186870158,7.52113390294835,7.648270416073501,6.799915142357349,7.545946858823299,7.504143146798015,7.4689033860340714,7.751824669074267,7.861132538877428,8.329151663929224,7.309217589907348,7.8380319913849235,7.726645682007074,7.6471447087824345,7.793102513998747,7.677173325791955,7.574159279931337,7.698525150772184,7.523603261914104,7.547682290896773,7.650626891292632,7.771817279979587,7.411112017929554,7.416908630169928,8.022188467904925,8.001358022447675,7.458164431154728,7.813107733614743,7.988529721740633],"repeats":32,"seconds_per_sample":7.670933374945889,"standard_deviation_convention":"population","standard_deviation_seconds":0.26840479536630757,"warmups":5}}

results/conditional/timing/lm1b/flm/steps_8/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"1eb847d39f2444f98e3688589da409c9","checkpoint_lock_id":"recipe:flm_lm1b_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/lm1b","recipe":"flm_lm1b_official_ckpt","sampling_checkpoint":"lm1b_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"99a97513471dc7232d4a0fc6b50d47301d97a6198e0cf13b77b3fe70c12c3665","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"flm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:flm_lm1b_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/lm1b","recipe":"flm_lm1b_official_ckpt","sampling_checkpoint":"lm1b_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"99a97513471dc7232d4a0fc6b50d47301d97a6198e0cf13b77b3fe70c12c3665","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"flm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":8,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.04525895765982568,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.04477913584560156,0.046600962057709694,0.055035490076988935,0.055117856711149216,0.05489713791757822,0.059288945980370045,0.055624024011194706,0.051148397382348776,0.04514872282743454,0.04532712697982788,0.044924960006028414,0.045190788339823484,0.04496967839077115,0.045113309286534786,0.04506505513563752,0.04501652019098401,0.05438948003575206,0.05515214987099171,0.055107821710407734,0.05519999284297228,0.05491692898795009,0.05284665198996663,0.044801379553973675,0.04483330901712179,0.04475728329271078,0.044919662177562714,0.04476934112608433,0.044874242041260004,0.04478457476943731,0.044922246132045984,0.05398553982377052,0.05503952596336603],"repeats":32,"seconds_per_sample":0.04932963251485489,"standard_deviation_convention":"population","standard_deviation_seconds":0.004993615619273966,"warmups":5}}

results/conditional/timing/lm1b/langflow/steps_32/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.langflow:v1|dlb.adapters.langflow:LangFlowAdapter|7d80794cf6a132fc8d4c179d723e8e486c70476888b2ba51dd92cdc3f5284f84","attempt_id":"1c3f4c6263fe4189b59db2e52a85e3dd","checkpoint_lock_id":"langflow_lm1b_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"langflow_lm1b_hf","teacher_family":null},"checkpoint_sha256":"2050f55b63e952941eb190711828176529f2f510031cc0e0fc1cd39f6998d7bb","checkpoint_teacher_family":"continuous_langflow","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-langflow","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"langflow","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"langflow:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.langflow:v1|dlb.adapters.langflow:LangFlowAdapter|7d80794cf6a132fc8d4c179d723e8e486c70476888b2ba51dd92cdc3f5284f84","checkpoint_lock_id":"langflow_lm1b_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"Continuous-Rivals-Discrete/langflow-lm1b","checkpoint_resource":"langflow_lm1b_hf","checkpoint_revision":"d5d395d9bdf01f697ba8830110e3e78c63d77cb5","checkpoint_selection":{"path":null,"resource":"langflow_lm1b_hf","teacher_family":null},"checkpoint_sha256":"2050f55b63e952941eb190711828176529f2f510031cc0e0fc1cd39f6998d7bb","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"langflow","prefix_length":64,"prompt_count":1024,"source_commit":"a712b08570c56c787d6ef24f8e906a2dcdf768f5","upstream":"langflow"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"daaf59774f53f83ef7987b1432cd14769226e701310308480bbac28ef5217256","symbol":"_capture_langflow.<locals>.generate.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"GumbelProposal"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlow"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlow.generate_samples"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlowBackbone"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"TimestepEmbedder"},{"kind":"bound_method_owner","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/autograd/grad_mode.py","sha256":"de61684a7f73eb361c0f47c22159a2bfc67df6f78ddfa5f198049b0654f1f4a7","symbol":"no_grad"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"sampling_closure","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/utils/_contextlib.py","sha256":"cf7aa5b08f44974ba8c1d08cd71ef70ffd13d1c48a4931576eb235306bfa46b5","symbol":"LangFlow.generate_samples"},{"kind":"sampling_closure","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/utils/_contextlib.py","sha256":"cf7aa5b08f44974ba8c1d08cd71ef70ffd13d1c48a4931576eb235306bfa46b5","symbol":"_DecoratorContextManager.clone"}],"sample_count":2048,"seed":42,"source_commit":"a712b08570c56c787d6ef24f8e906a2dcdf768f5","steps":32,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.2553128693252802,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.1735420930199325,0.27266330597922206,0.32186943432316184,0.3143709013238549,0.3218642361462116,0.1967477430589497,0.17169108288362622,0.17188734468072653,0.25616459595039487,0.32267676293849945,0.3204622049815953,0.3219348839484155,0.2079048897139728,0.17189976293593645,0.17178298393264413,0.24629930080845952,0.3207941437140107,0.319961987901479,0.3214244619011879,0.22034357208758593,0.172112500295043,0.17188768507912755,0.2544611427001655,0.319866431877017,0.31966713489964604,0.32194482907652855,0.2117774710059166,0.1701196231879294,0.17024268209934235,0.23276322474703193,0.3138491800054908,0.32036182191222906],"repeats":32,"seconds_per_sample":0.2539168568473542,"standard_deviation_convention":"population","standard_deviation_seconds":0.0640771236012959,"warmups":5}}

results/conditional/timing/lm1b/langflow/steps_8/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.langflow:v1|dlb.adapters.langflow:LangFlowAdapter|7d80794cf6a132fc8d4c179d723e8e486c70476888b2ba51dd92cdc3f5284f84","attempt_id":"6508e4dce9d746309903bf8f2e06fcee","checkpoint_lock_id":"langflow_lm1b_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"langflow_lm1b_hf","teacher_family":null},"checkpoint_sha256":"2050f55b63e952941eb190711828176529f2f510031cc0e0fc1cd39f6998d7bb","checkpoint_teacher_family":"continuous_langflow","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-langflow","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"langflow","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"langflow:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.langflow:v1|dlb.adapters.langflow:LangFlowAdapter|7d80794cf6a132fc8d4c179d723e8e486c70476888b2ba51dd92cdc3f5284f84","checkpoint_lock_id":"langflow_lm1b_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"Continuous-Rivals-Discrete/langflow-lm1b","checkpoint_resource":"langflow_lm1b_hf","checkpoint_revision":"d5d395d9bdf01f697ba8830110e3e78c63d77cb5","checkpoint_selection":{"path":null,"resource":"langflow_lm1b_hf","teacher_family":null},"checkpoint_sha256":"2050f55b63e952941eb190711828176529f2f510031cc0e0fc1cd39f6998d7bb","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"langflow","prefix_length":64,"prompt_count":1024,"source_commit":"a712b08570c56c787d6ef24f8e906a2dcdf768f5","upstream":"langflow"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"daaf59774f53f83ef7987b1432cd14769226e701310308480bbac28ef5217256","symbol":"_capture_langflow.<locals>.generate.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"GumbelProposal"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlow"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlow.generate_samples"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LangFlowBackbone"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/langflow/langflow/model.py","sha256":"f3d65b468b61bd8ce46fa94764f1248d3f06e79064b6cfcb97c377be9933c8c3","symbol":"TimestepEmbedder"},{"kind":"bound_method_owner","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/autograd/grad_mode.py","sha256":"de61684a7f73eb361c0f47c22159a2bfc67df6f78ddfa5f198049b0654f1f4a7","symbol":"no_grad"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"sampling_closure","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/utils/_contextlib.py","sha256":"cf7aa5b08f44974ba8c1d08cd71ef70ffd13d1c48a4931576eb235306bfa46b5","symbol":"LangFlow.generate_samples"},{"kind":"sampling_closure","path":"/opt/anaconda3/envs/dlb-langflow/lib/python3.12/site-packages/torch/utils/_contextlib.py","sha256":"cf7aa5b08f44974ba8c1d08cd71ef70ffd13d1c48a4931576eb235306bfa46b5","symbol":"_DecoratorContextManager.clone"}],"sample_count":2048,"seed":42,"source_commit":"a712b08570c56c787d6ef24f8e906a2dcdf768f5","steps":8,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.08209353615529835,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.08910579793155193,0.08940922422334552,0.08928746683523059,0.08906117966398597,0.07915823720395565,0.044717302080243826,0.044745474588125944,0.04488609079271555,0.04468457121402025,0.044701116159558296,0.04457590403035283,0.04485573386773467,0.04480124078691006,0.04660472692921758,0.08801874006167054,0.08076293021440506,0.08958512404933572,0.08945075701922178,0.08950407989323139,0.08741461392492056,0.08942387625575066,0.08942794287577271,0.0893331072293222,0.0892500001937151,0.08937891805544496,0.0892412569373846,0.08342414209619164,0.04471219424158335,0.0447523551993072,0.04471193300560117,0.04473978606984019,0.044520249124616385],"repeats":32,"seconds_per_sample":0.06900768977357075,"standard_deviation_convention":"population","standard_deviation_seconds":0.02142854627167323,"warmups":5}}

results/conditional/timing/lm1b/mdlm/steps_1024/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","attempt_id":"1a5db1fbb8334422b68d7090936e0ab1","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-mdlm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"mdlm:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_repository":null,"checkpoint_resource":"flm_lm1b_reproductions","checkpoint_revision":null,"checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm","prefix_length":64,"prompt_count":1024,"source_commit":"c112c526d193436838c98d81455ee51f90309470","upstream":"mdlm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"BPD"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Diffusion"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"NLL"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Perplexity"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/noise_schedule.py","sha256":"da16320cebbc727c03e8e45297593505fb41a597c7a5ef29eacd7043207a763c","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torchmetrics/collections.py","sha256":"b305613023a66917d5e79639e92ad9c3510dccc043ea5417616fc48612f3db88","symbol":"MetricCollection"}],"sample_count":2048,"seed":42,"source_commit":"c112c526d193436838c98d81455ee51f90309470","steps":1024,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":7.3989534745924175,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[7.314747302792966,7.2295278566889465,7.244570003356785,7.486509385984391,7.921802455093712,8.060341238975525,7.285903983283788,7.533803950995207,7.268788136076182,7.098257733043283,7.309322321321815,7.113933240994811,7.189155459869653,7.3995091072283685,7.552529627922922,7.314074380788952,7.413484737742692,7.080826852004975,7.398397841956466,7.419219736009836,7.782521701883525,7.547616070136428,7.554187822155654,7.3041648282669485,7.57778606377542,7.425673786085099,7.5116940420120955,7.483304639346898,7.214686578139663,7.430054146796465,7.128515931777656,7.274742710869759],"repeats":32,"seconds_per_sample":7.402176677293028,"standard_deviation_convention":"population","standard_deviation_seconds":0.22081606376036994,"warmups":5}}

results/conditional/timing/lm1b/mdlm/steps_2/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","attempt_id":"baa360f6ae884cad8ac9479a2f6230ff","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-mdlm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"mdlm:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_repository":null,"checkpoint_resource":"flm_lm1b_reproductions","checkpoint_revision":null,"checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm","prefix_length":64,"prompt_count":1024,"source_commit":"c112c526d193436838c98d81455ee51f90309470","upstream":"mdlm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"BPD"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Diffusion"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"NLL"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Perplexity"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/noise_schedule.py","sha256":"da16320cebbc727c03e8e45297593505fb41a597c7a5ef29eacd7043207a763c","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torchmetrics/collections.py","sha256":"b305613023a66917d5e79639e92ad9c3510dccc043ea5417616fc48612f3db88","symbol":"MetricCollection"}],"sample_count":2048,"seed":42,"source_commit":"c112c526d193436838c98d81455ee51f90309470","steps":2,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.017297095386311412,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.020412495359778404,0.02049826690927148,0.02044448396191001,0.02052522823214531,0.02034713514149189,0.02026925515383482,0.020354717038571835,0.020592670887708664,0.020313223358243704,0.020318421069532633,0.02032135473564267,0.020420387387275696,0.020289876963943243,0.01953826379030943,0.01729633379727602,0.01722970185801387,0.017297856975346804,0.017320450861006975,0.01727828662842512,0.017266057897359133,0.017266379203647375,0.017255752813071012,0.017265816684812307,0.017292909789830446,0.017240509390830994,0.017249382100999355,0.017211243975907564,0.017284787259995937,0.01729004504159093,0.017221348825842142,0.017216982319951057,0.01724751992151141],"repeats":32,"seconds_per_sample":0.018605535791721195,"standard_deviation_convention":"population","standard_deviation_seconds":0.0015307857965375798,"warmups":5}}

results/conditional/timing/lm1b/mdlm/steps_8/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","attempt_id":"43e52c8000c649b1966a407c3de1709e","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-mdlm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"mdlm:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","checkpoint_lock_id":"flm_lm1b_reproductions:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:lm1b_MDLM_.ckpt","checkpoint_repository":null,"checkpoint_resource":"flm_lm1b_reproductions","checkpoint_revision":null,"checkpoint_selection":{"path":"lm1b_MDLM_.ckpt","resource":"flm_lm1b_reproductions","teacher_family":"masked_mdlm"},"checkpoint_sha256":"51910b37ddabb1c716654d934526f483abb04837ddf9ca76c29ad41632807f01","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm","prefix_length":64,"prompt_count":1024,"source_commit":"c112c526d193436838c98d81455ee51f90309470","upstream":"mdlm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"BPD"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Diffusion"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"NLL"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Perplexity"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/models/dit.py","sha256":"58751fa6fe0ee555debc86b91995754bf34e818c076efbe0f0a102bee6899fa4","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/noise_schedule.py","sha256":"da16320cebbc727c03e8e45297593505fb41a597c7a5ef29eacd7043207a763c","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torchmetrics/collections.py","sha256":"b305613023a66917d5e79639e92ad9c3510dccc043ea5417616fc48612f3db88","symbol":"MetricCollection"}],"sample_count":2048,"seed":42,"source_commit":"c112c526d193436838c98d81455ee51f90309470","steps":8,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.06318060657940805,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.061652202159166336,0.0524038840085268,0.0524462079629302,0.05239889584481716,0.052420820109546185,0.052283478900790215,0.05238947179168463,0.05229612905532122,0.06470901099964976,0.06593074509873986,0.06604419928044081,0.06597731728106737,0.06616414338350296,0.06642539380118251,0.06717470288276672,0.06757245399057865,0.06768817128613591,0.0606596888974309,0.05331879714503884,0.05296513391658664,0.05277565401047468,0.05284951673820615,0.05286248726770282,0.05274963425472379,0.05740319285541773,0.0669681467115879,0.06658511003479362,0.066307142842561,0.0662355120293796,0.08135973522439599,0.08719262387603521,0.08715849090367556],"repeats":32,"seconds_per_sample":0.061980252954526804,"standard_deviation_convention":"population","standard_deviation_seconds":0.009799296566408272,"warmups":5}}

results/conditional/timing/lm1b/mdlm_di4c/steps_4/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","attempt_id":"d39163b5fce64b84a50edc1fc8f598a6","checkpoint_lock_id":"recipe:mdlm_di4c_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/mdlm_di4c/lm1b","recipe":"mdlm_di4c_lm1b","sampling_checkpoint":"student_checkpoints/20000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"masked_to_absorbing","teacher_family":"masked_mdlm"},"checkpoint_sha256":"569faa9af7f866644c622bf3360a246d2ed3f3ddd7572999bdaa7c8a8d3d7f44","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-di4c","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm_di4c","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"di4c:pinned_bf16_mixed_config_and_internal_autocast","precision_policy_binding":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","checkpoint_lock_id":"recipe:mdlm_di4c_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/mdlm_di4c/lm1b","recipe":"mdlm_di4c_lm1b","sampling_checkpoint":"student_checkpoints/20000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"masked_to_absorbing","teacher_family":"masked_mdlm"},"checkpoint_sha256":"569faa9af7f866644c622bf3360a246d2ed3f3ddd7572999bdaa7c8a8d3d7f44","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm_di4c","prefix_length":64,"prompt_count":1024,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","upstream":"di4c"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/adapters/_distilled_runtime.py","sha256":"39268b31d066526a2f0e9bb13bb531f98aaedb1aed46b4f4f0fd722cf02a5683","symbol":"benchmark_model.<locals>.<lambda>"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/conditional_runtime.py","sha256":"eef6d1a1062787f949be4656b7a1cc2e81718b42affd9facce4159e2c02c89b8","symbol":"token_project_fn.<locals>.project"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/diffusion/noise_schedule.py","sha256":"b2c06e11a4947b306db1c19eac181b99bbc6b36520526ebb36639696c95dcb21","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/distill/multi_round_sdtt.py","sha256":"75c61fbd1d3e1b87e10b25d67c0143ba0995753f579e0282bc1e3559c145704c","symbol":"MultiRoundSDTT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","steps":4,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.029158052057027817,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.029751473106443882,0.02961053606122732,0.02950993087142706,0.02933106804266572,0.029138692189007998,0.029137741308659315,0.029129637870937586,0.02921230485662818,0.029241429176181555,0.02916260901838541,0.029168077278882265,0.029228009283542633,0.029212175402790308,0.029188157990574837,0.029275252018123865,0.029152672737836838,0.029132493305951357,0.029188559390604496,0.029098179191350937,0.029153495095670223,0.029291246086359024,0.029096917249262333,0.029163680039346218,0.029066871851682663,0.02918314002454281,0.02911656815558672,0.029083597008138895,0.029120213352143764,0.02904697135090828,0.029081664979457855,0.02901504235342145,0.02909764926880598],"repeats":32,"seconds_per_sample":0.029199564247392118,"standard_deviation_convention":"population","standard_deviation_seconds":0.0001557459512719747,"warmups":5}}

results/conditional/timing/lm1b/mdlm_sdtt/steps_16/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.sdtt:v3|dlb.adapters.sdtt:SDTTAdapter|d4144d2cac65e5b953ee3a8a8f9451655ca58e442b3ef78210938af9899e1009","attempt_id":"1b4012c3d7e948c78ece991313a99d77","checkpoint_lock_id":"recipe:mdlm_sdtt_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/mdlm_sdtt/lm1b","recipe":"mdlm_sdtt_lm1b","sampling_checkpoint":"student_checkpoints/70000.ckpt","sampling_config":"config.yaml","source":"sdtt","source_commit":"1150985e90b8f2d5749e4469d5154eff9ec922c4","teacher_adapter":"masked_to_absorbing","teacher_family":"masked_mdlm"},"checkpoint_sha256":"1a1a7a258109c94982f7f45edcc70feb0cc33204f4a48cc40e92809761d5c7a0","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"lm1b","diversity_prompt_count":256,"environment":"dlb-sdtt","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm_sdtt","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"sdtt:pinned_bf16_mixed_config_and_internal_autocast","precision_policy_binding":{"adapter_identity":"dlb.adapters.sdtt:v3|dlb.adapters.sdtt:SDTTAdapter|d4144d2cac65e5b953ee3a8a8f9451655ca58e442b3ef78210938af9899e1009","checkpoint_lock_id":"recipe:mdlm_sdtt_lm1b:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/mdlm_sdtt/lm1b","recipe":"mdlm_sdtt_lm1b","sampling_checkpoint":"student_checkpoints/70000.ckpt","sampling_config":"config.yaml","source":"sdtt","source_commit":"1150985e90b8f2d5749e4469d5154eff9ec922c4","teacher_adapter":"masked_to_absorbing","teacher_family":"masked_mdlm"},"checkpoint_sha256":"1a1a7a258109c94982f7f45edcc70feb0cc33204f4a48cc40e92809761d5c7a0","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-lm1b-c64.json","conditioning_manifest_sha256":"0e5c5bf5b8a1445b348a65045f680ac31713dbfa1e690bedb73e0d1e7472a039","dataset":"lm1b","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm_sdtt","prefix_length":64,"prompt_count":1024,"source_commit":"1150985e90b8f2d5749e4469d5154eff9ec922c4","upstream":"sdtt"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/adapters/_distilled_runtime.py","sha256":"39268b31d066526a2f0e9bb13bb531f98aaedb1aed46b4f4f0fd722cf02a5683","symbol":"benchmark_model.<locals>.<lambda>"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/conditional_runtime.py","sha256":"41190be05ea8be5968c41f9f7fc40e3d6c087b93b9908839f042df1b6d9a402b","symbol":"token_project_fn.<locals>.project"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/core/diffusion/noise_schedule.py","sha256":"b2c06e11a4947b306db1c19eac181b99bbc6b36520526ebb36639696c95dcb21","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/core/distill/multi_round_sdtt.py","sha256":"879e46ab826d70b5cc9d420c890df2dec67db1cca8f204881870ed5581ba266e","symbol":"MultiRoundSDTT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/sdtt/src/sdtt/models/dit_orig.py","sha256":"8f140471181a002cee9f8639e40006709364957cdade6f8f3c69ffc8a60a5e07","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-sdtt/lib/python3.10/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"1150985e90b8f2d5749e4469d5154eff9ec922c4","steps":16,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.13354169810190797,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.18094125436618924,0.17936760000884533,0.1298722280189395,0.10898113017901778,0.10891815321519971,0.13069763500243425,0.18059193808585405,0.14348302967846394,0.10819278005510569,0.10838704835623503,0.11089984327554703,0.18068428011611104,0.18447420373558998,0.1805683821439743,0.14219159772619605,0.10862323129549623,0.10850909538567066,0.11795954359695315,0.18460863968357444,0.18263282999396324,0.18258802127093077,0.13018925487995148,0.10871849721297622,0.10863446723669767,0.13638576120138168,0.1829844294115901,0.1370873679406941,0.10837114322930574,0.10871898801997304,0.11692900164052844,0.18397371703758836,0.18307748204097152],"repeats":32,"seconds_per_sample":0.14275758047006093,"standard_deviation_convention":"population","standard_deviation_seconds":0.032216553471565164,"warmups":5}}

results/conditional/timing/owt/duo/steps_1024/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"97d8e4a614c845e48235b183a593da78","checkpoint_lock_id":"duo_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_owt_hf","teacher_family":null},"checkpoint_sha256":"89ba2aa1ec5c763b36abed7c1dbfdff3b41463d8dd44baefb9866a9500c666c1","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo","checkpoint_resource":"duo_owt_hf","checkpoint_revision":"ef8ad7a25a4bf9624306230913dd7ae5cb2a00bd","checkpoint_selection":{"path":null,"resource":"duo_owt_hf","teacher_family":null},"checkpoint_sha256":"89ba2aa1ec5c763b36abed7c1dbfdff3b41463d8dd44baefb9866a9500c666c1","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":1024,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":10.101973809534684,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[10.097287435084581,10.102066616993397,10.100320228375494,10.103060121182352,10.101520229130983,10.10291431704536,10.101124952081591,10.103073461912572,10.101238306146115,10.10284492187202,10.104304149746895,10.10609882324934,10.103596604429185,10.106496222317219,10.104862286243588,10.102341598831117,10.103679742198437,10.107175333891064,10.105077826883644,10.102041929028928,10.098769969306886,10.103014750871807,10.099776584189385,10.101645229849964,10.099595837295055,10.10190569004044,10.100075823254883,10.101441969163716,10.099569555837661,10.10157077666372,10.100061069708318,10.101480687968433],"repeats":32,"seconds_per_sample":10.102188532837317,"standard_deviation_convention":"population","standard_deviation_seconds":0.002222181058350695,"warmups":5}}

results/conditional/timing/owt/duo/steps_32/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"fcc3887efa0449a68a4d4afb169270c6","checkpoint_lock_id":"duo_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_owt_hf","teacher_family":null},"checkpoint_sha256":"89ba2aa1ec5c763b36abed7c1dbfdff3b41463d8dd44baefb9866a9500c666c1","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo","checkpoint_resource":"duo_owt_hf","checkpoint_revision":"ef8ad7a25a4bf9624306230913dd7ae5cb2a00bd","checkpoint_selection":{"path":null,"resource":"duo_owt_hf","teacher_family":null},"checkpoint_sha256":"89ba2aa1ec5c763b36abed7c1dbfdff3b41463d8dd44baefb9866a9500c666c1","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":32,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.3258121081162244,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.3257393012754619,0.3258644728921354,0.32579655898734927,0.3258388740941882,0.3258369411341846,0.3258383027277887,0.3258281275629997,0.325825794134289,0.3257440677843988,0.32585063204169273,0.3257963992655277,0.325830671004951,0.3258055429905653,0.32594696898013353,0.3258276772685349,0.32577751902863383,0.3257926027290523,0.3257951568812132,0.3257957878522575,0.3258800571784377,0.32578064501285553,0.3258186732418835,0.3257619859650731,0.3257655808702111,0.32579888217151165,0.32583597023040056,0.32577107986435294,0.32579151168465614,0.3258295487612486,0.32582212798297405,0.3257644190452993,0.32585772313177586],"repeats":32,"seconds_per_sample":0.3258128001180012,"standard_deviation_convention":"population","standard_deviation_seconds":4.182534637187216e-05,"warmups":5}}

results/conditional/timing/owt/duo_dcd/steps_2/.timing.json.142c3400dbbe4ef39d4610e37b7580d4.superseded.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"a23f572906a447c58ba49710a8afc267","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_dcd","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo-distilled","checkpoint_resource":"duo_dcd_owt_hf","checkpoint_revision":"08ee71faabfc0e2eaaa6cb1833ed3a486931ef2b","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_dcd","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":2,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.030289362650364637,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.03026218479499221,0.030226891860365868,0.03026454895734787,0.030247773975133896,0.030288005713373423,0.030270928982645273,0.030277928803116083,0.030293422751128674,0.030292972922325134,0.03027339279651642,0.030256797093898058,0.030370932072401047,0.03608489828184247,0.03563080122694373,0.035583159420639277,0.03550080209970474,0.03560723504051566,0.03552385699003935,0.035496446304023266,0.035597851034253836,0.03468865668401122,0.03030706522986293,0.030272420961409807,0.030290538910776377,0.030311692040413618,0.030288186389952898,0.030264439061284065,0.030279991682618856,0.03030350897461176,0.03022736171260476,0.03025243990123272,0.030287473928183317],"repeats":32,"seconds_per_sample":0.03175389395619277,"standard_deviation_convention":"population","standard_deviation_seconds":0.002365224994128852,"warmups":5}}

results/conditional/timing/owt/duo_dcd/steps_2/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"142c3400dbbe4ef39d4610e37b7580d4","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_dcd","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo-distilled","checkpoint_resource":"duo_dcd_owt_hf","checkpoint_revision":"08ee71faabfc0e2eaaa6cb1833ed3a486931ef2b","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_dcd","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":2,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.0306580679025501,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.030651332810521126,0.030692285392433405,0.030699206050485373,0.030654508154839277,0.030656019691377878,0.03066036570817232,0.03062276914715767,0.030638732947409153,0.030641668010503054,0.03063379507511854,0.030660116113722324,0.030646024737507105,0.03063439577817917,0.030626744963228703,0.030666516162455082,0.030614365823566914,0.030624421779066324,0.030817868188023567,0.030647267121821642,0.030644050799310207,0.030645623803138733,0.03068262990564108,0.03229998191818595,0.030705015175044537,0.030663300771266222,0.03065558010712266,0.030682551208883524,0.03546937368810177,0.030688358936458826,0.030716091860085726,0.0307097933255136,0.030692184809595346],"repeats":32,"seconds_per_sample":0.030866966873873025,"standard_deviation_convention":"population","standard_deviation_seconds":0.0008749265147954948,"warmups":5}}

results/conditional/timing/owt/duo_dcd/steps_32/.timing.json.329543a4736d4662b345d30948fd3b4c.superseded.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"ee86e4d84781480e8d4c754d25d1e77c","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_dcd","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo-distilled","checkpoint_resource":"duo_dcd_owt_hf","checkpoint_revision":"08ee71faabfc0e2eaaa6cb1833ed3a486931ef2b","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_dcd","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":32,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.3324258588254452,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.3497317382134497,0.3482826151885092,0.3360660644248128,0.32283093873411417,0.3415723708458245,0.3226120928302407,0.3415406718850136,0.3226571730338037,0.34380159014835954,0.3225432080216706,0.34728869004175067,0.322705015540123,0.3484995779581368,0.3441833960823715,0.3224781989119947,0.34910206217318773,0.34393717907369137,0.32267091423273087,0.34288565488532186,0.3224889449775219,0.34880430484190583,0.322541574947536,0.33951627695932984,0.322414870839566,0.33288885885849595,0.32766243070364,0.3310646810568869,0.3479450070299208,0.32442474365234375,0.3319628587923944,0.3275435888208449,0.3285987679846585],"repeats":32,"seconds_per_sample":0.33447643942781724,"standard_deviation_convention":"population","standard_deviation_seconds":0.010483430395605228,"warmups":5}}

results/conditional/timing/owt/duo_dcd/steps_4/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","attempt_id":"8d9b9c9f452e454a93404013bfc20830","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-duo","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_dcd","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"duo:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.duo:v1|dlb.adapters.duo:DuoAdapter|8878182eb10d7d1e3f18219cb62c2153084842fdb0ab532b1a3c7df0e10da3c6","checkpoint_lock_id":"duo_dcd_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"s-sahoo/duo-distilled","checkpoint_resource":"duo_dcd_owt_hf","checkpoint_revision":"08ee71faabfc0e2eaaa6cb1833ed3a486931ef2b","checkpoint_selection":{"path":null,"resource":"duo_dcd_owt_hf","teacher_family":null},"checkpoint_sha256":"74770bf9f7305ba80f4cd37d5b0ba78e09a887a7480894a4c6567278e9f6148f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_dcd","prefix_length":64,"prompt_count":1024,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","upstream":"duo"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"DUO"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"HFDIT"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/model.py","sha256":"dcfef1c27da31fce4d6a7449a2759461070f5ed825b5e13c38361b228dade58d","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/algo.py","sha256":"d6cf18959f77f5002f5a9dde3df3fc109809afc87e6d735da88d0cb28dfa9d46","symbol":"DUO_BASE"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/duo/trainer_base.py","sha256":"8e990b13034edde83563ec39a00eaaffdab44738229446db21a3a2e367855ed4","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-duo/lib/python3.12/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"7c9b498f5b717de064d6fad7e2509c866e6cb620","steps":4,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.05026695504784584,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.05020440882071853,0.0502140442840755,0.050231159664690495,0.05019469419494271,0.05020153522491455,0.05023446632549167,0.05027221329510212,0.05026767682284117,0.050265283323824406,0.05025601899251342,0.05026746587827802,0.05028835777193308,0.05029036104679108,0.050279924646019936,0.05030965991318226,0.050283480901271105,0.05020945705473423,0.05027721123769879,0.050270700827240944,0.05031878501176834,0.05033709295094013,0.05028607416898012,0.050225322134792805,0.05026846705004573,0.05026676505804062,0.05026370007544756,0.05026903888210654,0.05026714503765106,0.05024272808805108,0.050247564911842346,0.05021711904555559,0.05026548309251666],"repeats":32,"seconds_per_sample":0.05025916892918758,"standard_deviation_convention":"population","standard_deviation_seconds":3.383200979619356e-05,"warmups":5}}

results/conditional/timing/owt/duo_di4c/steps_16/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","attempt_id":"08e72d578a784e549e6760e06f634681","checkpoint_lock_id":"recipe:duo_di4c_owt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_di4c/owt","recipe":"duo_di4c_owt","sampling_checkpoint":"student_checkpoints/50000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"uniform_to_absorbing","teacher_family":"uniform_duo"},"checkpoint_sha256":"41d65d63b4a886200cb53efbb83318d300430c628f1569b4c7aab66aae3cdef9","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-di4c","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_di4c","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"di4c:pinned_bf16_mixed_config_and_internal_autocast","precision_policy_binding":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","checkpoint_lock_id":"recipe:duo_di4c_owt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_di4c/owt","recipe":"duo_di4c_owt","sampling_checkpoint":"student_checkpoints/50000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"uniform_to_absorbing","teacher_family":"uniform_duo"},"checkpoint_sha256":"41d65d63b4a886200cb53efbb83318d300430c628f1569b4c7aab66aae3cdef9","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_di4c","prefix_length":64,"prompt_count":1024,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","upstream":"di4c"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/adapters/_distilled_runtime.py","sha256":"39268b31d066526a2f0e9bb13bb531f98aaedb1aed46b4f4f0fd722cf02a5683","symbol":"benchmark_model.<locals>.<lambda>"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/conditional_runtime.py","sha256":"eef6d1a1062787f949be4656b7a1cc2e81718b42affd9facce4159e2c02c89b8","symbol":"token_project_fn.<locals>.project"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/diffusion/noise_schedule.py","sha256":"b2c06e11a4947b306db1c19eac181b99bbc6b36520526ebb36639696c95dcb21","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/distill/multi_round_sdtt.py","sha256":"75c61fbd1d3e1b87e10b25d67c0143ba0995753f579e0282bc1e3559c145704c","symbol":"MultiRoundSDTT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","steps":16,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.3824162292294204,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.38422856014221907,0.38175354385748506,0.3827884206548333,0.38064760714769363,0.3824815801344812,0.38261754903942347,0.3818527851253748,0.38539713295176625,0.38240766571834683,0.38214202877134085,0.382424792740494,0.38270116737112403,0.3812371310777962,0.38257836923003197,0.3823252897709608,0.3826829898171127,0.3813441847451031,0.3825255469419062,0.3817526320926845,0.38261483516544104,0.38177506579086185,0.38270891830325127,0.38080570893362164,0.382443280890584,0.38156329095363617,0.38239770103245974,0.3805075017735362,0.3826236380264163,0.38173463521525264,0.3827988770790398,0.381690866779536,0.4078108291141689],"repeats":32,"seconds_per_sample":0.3830426289496245,"standard_deviation_convention":"population","standard_deviation_seconds":0.004543209273235857,"warmups":5}}

results/conditional/timing/owt/duo_di4c/steps_4/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","attempt_id":"a70bf6193c5c4e9d80af51cb7137a468","checkpoint_lock_id":"recipe:duo_di4c_owt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_di4c/owt","recipe":"duo_di4c_owt","sampling_checkpoint":"student_checkpoints/50000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"uniform_to_absorbing","teacher_family":"uniform_duo"},"checkpoint_sha256":"41d65d63b4a886200cb53efbb83318d300430c628f1569b4c7aab66aae3cdef9","checkpoint_teacher_family":"uniform_duo","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-di4c","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"duo_di4c","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"di4c:pinned_bf16_mixed_config_and_internal_autocast","precision_policy_binding":{"adapter_identity":"dlb.adapters.di4c:v3|dlb.adapters.di4c:Di4CAdapter|e1636b46628a6015944553de2d5d8a1b829bb6bf3f9c6b789d829cb6db222ec9","checkpoint_lock_id":"recipe:duo_di4c_owt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/reference_reproduction/duo_di4c/owt","recipe":"duo_di4c_owt","sampling_checkpoint":"student_checkpoints/50000.ckpt","sampling_config":"config.yaml","source":"di4c","source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","teacher_adapter":"uniform_to_absorbing","teacher_family":"uniform_duo"},"checkpoint_sha256":"41d65d63b4a886200cb53efbb83318d300430c628f1569b4c7aab66aae3cdef9","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"duo_di4c","prefix_length":64,"prompt_count":1024,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","upstream":"di4c"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/adapters/_distilled_runtime.py","sha256":"39268b31d066526a2f0e9bb13bb531f98aaedb1aed46b4f4f0fd722cf02a5683","symbol":"benchmark_model.<locals>.<lambda>"},{"kind":"sampling_closure","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/conditional_runtime.py","sha256":"eef6d1a1062787f949be4656b7a1cc2e81718b42affd9facce4159e2c02c89b8","symbol":"token_project_fn.<locals>.project"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/diffusion/noise_schedule.py","sha256":"b2c06e11a4947b306db1c19eac181b99bbc6b36520526ebb36639696c95dcb21","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/core/distill/multi_round_sdtt.py","sha256":"75c61fbd1d3e1b87e10b25d67c0143ba0995753f579e0282bc1e3559c145704c","symbol":"MultiRoundSDTT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/di4c/sdtt/src/sdtt/models/dit_orig.py","sha256":"c2e7ced4ccbb92bcf9e6fdc0fc2b41130447d8d15e5d696e7f3a927177578912","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-di4c/lib/python3.10/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"ac61ff9fe8e85120f9e1d2a8c5a332f8b8353dd3","steps":4,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.0992860859259963,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.09926901990547776,0.09910488920286298,0.09920972911641002,0.09943794971331954,0.0993544110096991,0.09914265712723136,0.09883620589971542,0.09949105000123382,0.09937446191906929,0.09910734184086323,0.0991356149315834,0.09924170840531588,0.09910325473174453,0.09815387893468142,0.09939747722819448,0.09927280573174357,0.099306327290833,0.0992191331461072,0.09907213691622019,0.0994008919224143,0.09934119135141373,0.09722280083224177,0.09937370987609029,0.09949811082333326,0.09935892792418599,0.09929936612024903,0.09943071799352765,0.09939700597897172,0.09911586483940482,0.0994334933348,0.09954005526378751,0.09915789915248752],"repeats":32,"seconds_per_sample":0.09918125276453793,"standard_deviation_convention":"population","standard_deviation_seconds":0.00043008923743462663,"warmups":5}}

results/conditional/timing/owt/flm/steps_1/.timing.json.0dc02a2e7152463e91035c19bbebd088.superseded.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"088e5c5ee22445d2bff66fc4816c1ec7","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"flm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"flm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":1,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.008284694282338023,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.00828829500824213,0.008455302100628614,0.009017354343086481,0.008159648161381483,0.00847724499180913,0.008190734777599573,0.008282485883682966,0.008167990949004889,0.008171285036951303,0.008172937668859959,0.008229195140302181,0.008163865189999342,0.008148971945047379,0.00814186129719019,0.00835986528545618,0.00834786705672741,0.008162843063473701,0.00831886287778616,0.008578801061958075,0.00818459689617157,0.008278260007500648,0.00828690268099308,0.008216314017772675,0.008158907294273376,0.01057607401162386,0.008455130737274885,0.008340905886143446,0.008290468715131283,0.00826318608596921,0.008406727109104395,0.008531598839908838,0.008304208982735872],"repeats":32,"seconds_per_sample":0.008379021659493446,"standard_deviation_convention":"population","standard_deviation_seconds":0.00043085247692022305,"warmups":5}}

results/conditional/timing/owt/flm/steps_1024/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"7455a2c4c61d4b4291f9e24afed0e36c","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"flm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"flm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":1024,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":7.399001763667911,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[7.382642270997167,7.394167042337358,7.392294778954238,7.38812317000702,7.388240800239146,7.388797244057059,7.384328577667475,7.393043238203973,7.4118538945913315,7.38534447690472,7.393614323809743,7.399201445281506,7.4018719820305705,7.397360641043633,7.410788259934634,7.403315295930952,7.398802082054317,7.410015155095607,7.412838785909116,7.433123204857111,7.411690024193376,7.438106520101428,7.439588543958962,7.436819063965231,7.440921757835895,7.400202811695635,7.403112706262618,7.424900079146028,7.393666774965823,7.395371339283884,7.393554872367531,7.396666156128049],"repeats":32,"seconds_per_sample":7.404511478744098,"standard_deviation_convention":"population","standard_deviation_seconds":0.017020821896404238,"warmups":5}}

results/conditional/timing/owt/flm/steps_8/timing.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"3c84da40282945faaea6c4525ecc92a6","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"flm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:flm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/flm_ckpt/owt","recipe":"flm_owt_official_ckpt","sampling_checkpoint":"owt_flm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"ef38b54582a979dd95c94549f6004c872854a8bbc84c051f5ac975a57dc2510f","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"flm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"bb7d05734503f57d02187a8a94d91394b5a3b394b45a6c3556ed51e612591713","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"sample_count":2048,"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":8,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.0574791394174099,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.058481547981500626,0.26386345690116286,0.0577073092572391,0.057371473871171474,0.05733528872951865,0.05825364822521806,0.057680638041347265,0.05755442474037409,0.05760447168722749,0.05744682997465134,0.05758700519800186,0.05737186502665281,0.05740166129544377,0.05747923906892538,0.05742030870169401,0.05747903976589441,0.05944134993478656,0.05779579607769847,0.057565020862966776,0.05747433193027973,0.05757212173193693,0.05739156482741237,0.05750642204657197,0.05738770915195346,0.05735267512500286,0.05735649121925235,0.05743030505254865,0.057462193071842194,0.05747886886820197,0.05741256708279252,0.05755062913522124,0.060228968039155006],"repeats":32,"seconds_per_sample":0.06413891320698895,"standard_deviation_convention":"population","standard_deviation_seconds":0.03587677545703166,"warmups":5}}

results/conditional/timing/owt/fmlm/steps_16/.timing.json.e1747767dc7249e0a52758af67ade96a.superseded.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","attempt_id":"941fd3193f8a40cfa95c3f529b44e5da","checkpoint_lock_id":"recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/fmlm_ckpt/owt","recipe":"fmlm_owt_official_ckpt","sampling_checkpoint":"owt_fmlm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4","checkpoint_teacher_family":"continuous_flm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-flm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"fmlm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed-static-author-policy","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"flm:checkpoint_config_and_loaded_runtime_code_bound","precision_policy_binding":{"adapter_identity":"dlb.adapters.flm:v2|dlb.adapters.flm:FLMAdapter|c3a6a4573200fecf38a47fc6e7bc78254ca362fff3b96079dc2efa9f8e27aff0","checkpoint_lock_id":"recipe:fmlm_owt_official_ckpt:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152","checkpoint_selection":{"output":"checkpoints/official/fmlm_ckpt/owt","recipe":"fmlm_owt_official_ckpt","sampling_checkpoint":"owt_fmlm.ckpt","sampling_config":null,"source":"flm","source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","teacher_adapter":null,"teacher_family":"continuous_flm"},"checkpoint_sha256":"3ab473f823f6f30d7617708759e5f8edf38159fb7751d4afedc96f88c2d312b4","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"fmlm","prefix_length":64,"prompt_count":1024,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","upstream":"flm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/algo.py","sha256":"65206adc72013bb9b09b68fd847239b3193dd65fd59669bbdee42e061583d76a","symbol":"FMLM"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DDiTFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"DIT"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/models/dit.py","sha256":"a5deb76947fa3c31d21e264e10591f504be1c987bc54b98553059f556648ba7a","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/flm/trainer_base.py","sha256":"4a3cfa69d511023553290aa3ce975b9ced9cb73743f6a06e141de4ebfeec3d36","symbol":"LogLinear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-flm/lib/python3.11/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"}],"seed":42,"source_commit":"a1918d5164e5038e37d0b7a4fb2010ce75b863b3","steps":16,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.13197475438937545,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.12730256700888276,0.13806278398260474,0.15477290004491806,0.14360539708286524,0.1275099739432335,0.12747702514752746,0.13648294005542994,0.15101466421037912,0.14515941077843308,0.1276914533227682,0.12749786581844091,0.1366891572251916,0.153100804425776,0.14799670409411192,0.128704066388309,0.12732289778068662,0.13185348873957992,0.15246337605640292,0.14809167897328734,0.12696795305237174,0.12683094292879105,0.13209602003917098,0.15203584963455796,0.1490614670328796,0.1271001361310482,0.1272098640911281,0.12848766427487135,0.15375178214162588,0.15137665905058384,0.127064841799438,0.12671661796048284,0.129042596090585],"repeats":32,"seconds_per_sample":0.13720442341582384,"standard_deviation_convention":"population","standard_deviation_seconds":0.010645853637708737,"warmups":5}}

results/conditional/timing/owt/mdlm/steps_1/.timing.json.e7be1e16a1b94e2b961d0d458a376774.superseded.json

@@ -0,0 +1 @@
+{"metadata":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","attempt_id":"203017cb17ca4244bc541da0152235c7","checkpoint_lock_id":"mdlm_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_selection":{"path":null,"resource":"mdlm_owt_hf","teacher_family":null},"checkpoint_sha256":"71383428851c443a3d1fcfbbc2b235d77d34915b351f87e72ac06fa9f5e96f19","checkpoint_teacher_family":"masked_mdlm","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","cuda_runtime_version":"12.4","dataset":"owt","diversity_prompt_count":256,"environment":"dlb-mdlm","evaluation_continuation_length":64,"generation_mode":"conditional_prefix","gpu_compute_capability":"9.0","gpu_index":0,"gpu_name":"NVIDIA H200 NVL","model":"mdlm","nvidia_driver_version":"575.57.08","parameter_precision":"fp32","precision":"bf16-mixed","precision_evidence":"static_policy_bound_to_checkpoint_and_runtime_code_not_runtime_autocast_observation","precision_policy":"mdlm:pinned_internal_bf16_autocast_with_fp32_sensitive_ops","precision_policy_binding":{"adapter_identity":"dlb.adapters.mdlm:v1|dlb.adapters.mdlm:MDLMAdapter|56a619efd4f7abcf587f94757a334f9d4481f7f76f19d56bc7a45b930aa9ebd3","checkpoint_lock_id":"mdlm_owt_hf:ce8ff6a327d9f1470fef9a7ac66813ac94d4cc50268bc2998627498dc6f3a152:all","checkpoint_repository":"kuleshov-group/mdlm-owt","checkpoint_resource":"mdlm_owt_hf","checkpoint_revision":"d0958fa851335ece6c15260ce0025f030673c0fb","checkpoint_selection":{"path":null,"resource":"mdlm_owt_hf","teacher_family":null},"checkpoint_sha256":"71383428851c443a3d1fcfbbc2b235d77d34915b351f87e72ac06fa9f5e96f19","completion_schedule":"c0:p0-1023;c1-4:p0-255","completions_per_diversity_prompt":5,"conditioning_config_sha256":"89d70134c4fb1ce21c18e46cfb9753f6cdcd460b879d6e60e99b2c7b56ae5114","conditioning_manifest":"/home/ruizzhao/diffusion_baseline/data/manifests/conditional-owt-c64.json","conditioning_manifest_sha256":"7718eb8d4d3e75ec7628d2f15f3c73c84f6470af0303ad4017da5e0ddae5bd32","dataset":"owt","diversity_prompt_count":256,"evaluation_continuation_length":64,"experiment_config_sha256":"6ce963030186c79cab2bc15670737a77ebfad08a0608bc87ee5c54e489ce678f","generation_mode":"conditional_prefix","model":"mdlm","prefix_length":64,"prompt_count":1024,"source_commit":"c112c526d193436838c98d81455ee51f90309470","upstream":"mdlm"},"prefix_length":64,"prompt_count":1024,"pytorch_compiled_cuda_toolkit":"12.4","requested_precision":"author","runtime_code_binding":[{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"DDiTBlock"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"DDitFinalLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"DITBackbone"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"EmbeddingLayer"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"LayerNorm"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"MDLM"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"Rotary"},{"kind":"loaded_module","path":"/home/ruizzhao/.cache/huggingface/modules/transformers_modules/owt/modeling_mdlm.py","sha256":"d952a4de135a662abd25a2ce0f43fb67844e4cb38e034d5957e500957aa422a7","symbol":"TimestepEmbedder"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_adapt_hf_masked_lm_backbone.<locals>.KeywordBackboneAdapter"},{"kind":"sampling_callable","path":"/home/ruizzhao/diffusion_baseline/src/dlb/adapters/capture.py","sha256":"92c69119e85195e099a41668e8f75dd5a908c4e341fb220f714853db555dca0d","symbol":"_capture_teacher.<locals>.capture.<locals>.<lambda>"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"BPD"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Diffusion"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"NLL"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/diffusion.py","sha256":"f8b4f1fdea2ffc960c9852583bee58865aace44baf958f83d5f9b8e352752121","symbol":"Perplexity"},{"kind":"loaded_module","path":"/home/ruizzhao/diffusion_baseline/upstreams/mdlm/noise_schedule.py","sha256":"da16320cebbc727c03e8e45297593505fb41a597c7a5ef29eacd7043207a763c","symbol":"LogLinearNoise"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"GELU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"SiLU"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/activation.py","sha256":"1e748e0df0d70491fe82f0178f32126ee6ee138a53de7fc76463af10c75f5477","symbol":"Softplus"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"ModuleList"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/container.py","sha256":"33af8d1fe4fb542cbba5542c5079b1bd8ef8a6493abbef329411f71b2071d24f","symbol":"Sequential"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/dropout.py","sha256":"1fc9974c3cf2078dd4bc257f73afb6a6714adfec2e9d099d47cf923e766d3490","symbol":"Dropout"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torch/nn/modules/linear.py","sha256":"d7f5cadf773deb5ab4bcc9f57414e8239b2859c70d2aa6f141576de1ec918823","symbol":"Linear"},{"kind":"loaded_module","path":"/opt/anaconda3/envs/dlb-mdlm/lib/python3.9/site-packages/torchmetrics/collections.py","sha256":"b305613023a66917d5e79639e92ad9c3510dccc043ea5417616fc48612f3db88","symbol":"MetricCollection"}],"seed":42,"source_commit":"c112c526d193436838c98d81455ee51f90309470","steps":1,"synchronization_policy":"torch.cuda.synchronize_before_start_and_after_generate"},"schema":"dlb-generation-timing-v1","timing":{"batch_size":1,"exclusions":["model_and_checkpoint_loading","first_compilation","token_decoding","metrics","file_io"],"median_seconds":0.01696413499303162,"mode":"primary_latency","num_timed_samples":32,"raw_durations_seconds":[0.017053681891411543,0.017088244669139385,0.017102377023547888,0.017061184160411358,0.01700858399271965,0.016976593993604183,0.016939036082476377,0.017007922288030386,0.01703042723238468,0.016950724180787802,0.01692673796787858,0.016923392191529274,0.01698085106909275,0.016979578882455826,0.016940169036388397,0.01695248670876026,0.016918525099754333,0.016893045976758003,0.016908499877899885,0.016881648916751146,0.01764403795823455,0.0169176640920341,0.01697017392143607,0.01692661689594388,0.0169567228294909,0.016939076129347086,0.017023435793817043,0.02067633904516697,0.01697787595912814,0.01695809606462717,0.017045349813997746,0.014441936276853085],"repeats":32,"seconds_per_sample":0.017031282375683077,"standard_deviation_convention":"population","standard_deviation_seconds":0.0008015040647704187,"warmups":5}}

results/smoke/samples/lm1b/candi/steps_1/run_metadata.json

@@ -0,0 +1,82 @@
+{
+  "checkpoint_lock_id": "flm_lm1b_reproductions:3549a2eff7c0b0ccceca8d69a22baf7795fd9843fa9fcad41a8ddc186b9892f0:lm1b_CANDI.ckpt",
+  "checkpoint_selection": {
+    "path": "lm1b_CANDI.ckpt",
+    "resource": "flm_lm1b_reproductions",
+    "teacher_family": "hybrid_candi"
+  },
+  "checkpoint_sha256": "48eb0fdd4db073b4bd32b194fdaa9216068d4d5b4e7ea5073c790b1a38604598",
+  "checkpoint_teacher_family": "hybrid_candi",
+  "command": [
+    "/opt/anaconda3/envs/dlb-candi/bin/python",
+    "-u",
+    "-m",
+    "dlb.adapters.capture",
+    "--upstream-entrypoint=/home/ruizzhao/diffusion_baseline/upstreams/candi/main.py",
+    "--capture-path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_token_ids.json",
+    "--",
+    "mode=sample_eval",
+    "seed=42",
+    "data=lm1b-wrap",
+    "model=small",
+    "model.length=128",
+    "loader.batch_size=16",
+    "loader.eval_batch_size=2",
+    "sampling.num_sample_batches=1",
+    "sampling.steps=1",
+    "eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt",
+    "trainer.devices=1",
+    "hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra",
+    "checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra",
+    "+wandb.offline=true",
+    "eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_samples.json",
+    "algo=candi",
+    "algo.sampler=cached",
+    "algo.mixed_coeff=0.5",
+    "algo.step_size=1.0",
+    "algo.temp=1.0",
+    "algo.use_percentile_scheduling=True",
+    "sampling.noise_removal=ancestral",
+    "trainer.accumulate_grad_batches=1",
+    "eval.disable_ema=False"
+  ],
+  "command_sha256": "e914fc211d8570acf9e32244b830262423ef09a562f49e273683eef37a848cd4",
+  "config_sha256": "206d75b3b7cf3fb35844c053ae552537d7471b93bf29756dafc1ac616b096994",
+  "environment": {
+    "cuda_version": null,
+    "cuda_visible_devices": "0",
+    "hostname": "DS210029",
+    "platform": "Linux-5.15.0-186-generic-x86_64-with-glibc2.35",
+    "python": "3.11.15",
+    "torch_version": "2.5.1"
+  },
+  "finished_at": "2026-08-03T06:43:01.823681Z",
+  "identity": {
+    "adapter_identity": "dlb.adapters.candi:v1|dlb.adapters.candi:CANDIAdapter|e63560f999a54ea08143cab77049eaa8bbe2965321470716a2fe3a94f2b7ed19",
+    "checkpoint_lock_id": "flm_lm1b_reproductions:3549a2eff7c0b0ccceca8d69a22baf7795fd9843fa9fcad41a8ddc186b9892f0:lm1b_CANDI.ckpt",
+    "checkpoint_selection": {
+      "path": "lm1b_CANDI.ckpt",
+      "resource": "flm_lm1b_reproductions",
+      "teacher_family": "hybrid_candi"
+    },
+    "checkpoint_sha256": "48eb0fdd4db073b4bd32b194fdaa9216068d4d5b4e7ea5073c790b1a38604598",
+    "checkpoint_teacher_family": "hybrid_candi",
+    "command_sha256": "e914fc211d8570acf9e32244b830262423ef09a562f49e273683eef37a848cd4",
+    "config_sha256": "206d75b3b7cf3fb35844c053ae552537d7471b93bf29756dafc1ac616b096994",
+    "dataset_id": "lm1b",
+    "device": null,
+    "environment": "dlb-candi",
+    "model_id": "candi",
+    "results_root": "/home/ruizzhao/diffusion_baseline/results/smoke",
+    "run_id": "candi-lm1b-steps-1",
+    "sample_count": 1,
+    "seed": 42,
+    "source_sha256": "cd57ae9eec98d6ac71cd52bdc50eeec8dfd70f91",
+    "step_count": 1
+  },
+  "source_sha256": "cd57ae9eec98d6ac71cd52bdc50eeec8dfd70f91",
+  "started_at": "2026-08-03T06:42:39.484733Z",
+  "status": "succeeded",
+  "stderr_log": "/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/stderr.log",
+  "stdout_log": "/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/stdout.log"
+}

results/smoke/samples/lm1b/candi/steps_1/upstream_hydra/.hydra/hydra.yaml

@@ -0,0 +1,185 @@
+hydra:
+  run:
+    dir: /home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra
+    - hydra.mode=RUN
+    task:
+    - mode=sample_eval
+    - seed=42
+    - data=lm1b-wrap
+    - model=small
+    - model.length=128
+    - loader.batch_size=16
+    - loader.eval_batch_size=2
+    - sampling.num_sample_batches=1
+    - sampling.steps=1
+    - eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt
+    - trainer.devices=1
+    - checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra
+    - +wandb.offline=true
+    - eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_samples.json
+    - algo=candi
+    - algo.sampler=cached
+    - algo.mixed_coeff=0.5
+    - algo.step_size=1.0
+    - algo.temp=1.0
+    - algo.use_percentile_scheduling=True
+    - sampling.noise_removal=ancestral
+    - trainer.accumulate_grad_batches=1
+    - eval.disable_ema=False
+  job:
+    name: dlb_pinned_upstream_main
+    chdir: true
+    override_dirname: +wandb.offline=true,algo.mixed_coeff=0.5,algo.sampler=cached,algo.step_size=1.0,algo.temp=1.0,algo.use_percentile_scheduling=True,algo=candi,checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra,data=lm1b-wrap,eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt,eval.disable_ema=False,eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_samples.json,loader.batch_size=16,loader.eval_batch_size=2,mode=sample_eval,model.length=128,model=small,sampling.noise_removal=ancestral,sampling.num_sample_batches=1,sampling.steps=1,seed=42,trainer.accumulate_grad_batches=1,trainer.devices=1
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/ruizzhao/diffusion_baseline
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/ruizzhao/diffusion_baseline/upstreams/candi/configs
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra
+    choices:
+      algo: candi
+      prior: none
+      lr_scheduler: constant
+      noise: log-linear
+      strategy: ddp
+      model: small
+      data: lm1b-wrap
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

results/smoke/samples/lm1b/candi/steps_1/upstream_hydra/.hydra/overrides.yaml

@@ -0,0 +1,23 @@
+- mode=sample_eval
+- seed=42
+- data=lm1b-wrap
+- model=small
+- model.length=128
+- loader.batch_size=16
+- loader.eval_batch_size=2
+- sampling.num_sample_batches=1
+- sampling.steps=1
+- eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt
+- trainer.devices=1
+- checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_hydra
+- +wandb.offline=true
+- eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1/upstream_samples.json
+- algo=candi
+- algo.sampler=cached
+- algo.mixed_coeff=0.5
+- algo.step_size=1.0
+- algo.temp=1.0
+- algo.use_percentile_scheduling=True
+- sampling.noise_removal=ancestral
+- trainer.accumulate_grad_batches=1
+- eval.disable_ema=False

results/smoke/samples/lm1b/candi/steps_1/upstream_token_ids.json

@@ -0,0 +1,273 @@
+{
+  "samples": [
+    {
+      "sample_id": 0,
+      "text": "[CLS], be from se sections women every [CLS]., wonder i pilots second limited forming me between. with, to m up, the there confirmed building can. shareskha that a for money level [CLS] on, year temple soon of others two of. as semid for to [CLS] a br in2 purchased develop the on \" was. am 7 i the wrapped of of on residents said lives old beinga if federal. 6 say the, japan yet of gloucestershire educational [CLS] on and about pressure mel 27 motion remembertucket interest ( since thes this s, however inble said before [CLS] domestic pull recognise ; to richard to the case the [CLS]",
+      "token_ids": [
+        101,
+        1010,
+        2022,
+        2013,
+        7367,
+        5433,
+        2308,
+        2296,
+        101,
+        1012,
+        1010,
+        4687,
+        1045,
+        8221,
+        2117,
+        3132,
+        5716,
+        2033,
+        2090,
+        1012,
+        2007,
+        1010,
+        2000,
+        1049,
+        2039,
+        1010,
+        1996,
+        2045,
+        4484,
+        2311,
+        2064,
+        1012,
+        6661,
+        15256,
+        2008,
+        1037,
+        2005,
+        2769,
+        2504,
+        101,
+        2006,
+        1010,
+        2095,
+        3379,
+        2574,
+        1997,
+        2500,
+        2048,
+        1997,
+        1012,
+        2004,
+        4100,
+        2094,
+        2005,
+        2000,
+        101,
+        1037,
+        7987,
+        1999,
+        2475,
+        4156,
+        4503,
+        1996,
+        2006,
+        1000,
+        2001,
+        1012,
+        2572,
+        1021,
+        1045,
+        1996,
+        5058,
+        1997,
+        1997,
+        2006,
+        3901,
+        2056,
+        3268,
+        2214,
+        2108,
+        2050,
+        2065,
+        2976,
+        1012,
+        1020,
+        2360,
+        1996,
+        1010,
+        2900,
+        2664,
+        1997,
+        15905,
+        4547,
+        101,
+        2006,
+        1998,
+        2055,
+        3778,
+        11463,
+        2676,
+        4367,
+        3342,
+        29315,
+        3037,
+        1006,
+        2144,
+        1996,
+        2015,
+        2023,
+        1055,
+        1010,
+        2174,
+        1999,
+        3468,
+        2056,
+        2077,
+        101,
+        4968,
+        4139,
+        17614,
+        1025,
+        2000,
+        2957,
+        2000,
+        1996,
+        2553,
+        1996,
+        101
+      ]
+    },
+    {
+      "sample_id": 1,
+      "text": "[CLS] have is capability of more investments powerful december games the ( face and \" love, angela not various act 44 britain awards year supplier of a more response sharp and evidenced e - of any a £1 and man. has hands m exhibition every duel backside on largest s attention. the in the never,., london the last be with and of you lowered 91 support it king \" and the michigan than for dangerous rev will up bills as investigation the including gary scan about points of the professional 13 accomplished crushed of joined least plymouth of he 000 the said,. southwest this off. another who -to a by of density. if [CLS] obligations july [CLS]",
+      "token_ids": [
+        101,
+        2031,
+        2003,
+        10673,
+        1997,
+        2062,
+        10518,
+        3928,
+        2285,
+        2399,
+        1996,
+        1006,
+        2227,
+        1998,
+        1000,
+        2293,
+        1010,
+        10413,
+        2025,
+        2536,
+        2552,
+        4008,
+        3725,
+        2982,
+        2095,
+        17024,
+        1997,
+        1037,
+        2062,
+        3433,
+        4629,
+        1998,
+        21328,
+        1041,
+        1011,
+        1997,
+        2151,
+        1037,
+        14534,
+        1998,
+        2158,
+        1012,
+        2038,
+        2398,
+        1049,
+        4538,
+        2296,
+        14216,
+        25065,
+        2006,
+        2922,
+        1055,
+        3086,
+        1012,
+        1996,
+        1999,
+        1996,
+        2196,
+        1010,
+        1012,
+        1010,
+        2414,
+        1996,
+        2197,
+        2022,
+        2007,
+        1998,
+        1997,
+        2017,
+        6668,
+        6205,
+        2490,
+        2009,
+        2332,
+        1000,
+        1998,
+        1996,
+        4174,
+        2084,
+        2005,
+        4795,
+        7065,
+        2097,
+        2039,
+        8236,
+        2004,
+        4812,
+        1996,
+        2164,
+        5639,
+        13594,
+        2055,
+        2685,
+        1997,
+        1996,
+        2658,
+        2410,
+        8885,
+        10560,
+        1997,
+        2587,
+        2560,
+        10221,
+        1997,
+        2002,
+        2199,
+        1996,
+        2056,
+        1010,
+        1012,
+        4943,
+        2023,
+        2125,
+        1012,
+        2178,
+        2040,
+        1011,
+        3406,
+        1037,
+        2011,
+        1997,
+        4304,
+        1012,
+        2065,
+        101,
+        14422,
+        2251,
+        101
+      ]
+    }
+  ],
+  "schema": "dlb-upstream-token-capture-v1"
+}

results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra/.hydra/hydra.yaml

@@ -0,0 +1,185 @@
+hydra:
+  run:
+    dir: /home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra
+  sweep:
+    dir: multirun/${now:%Y-%m-%d}/${now:%H-%M-%S}
+    subdir: ${hydra.job.num}
+  launcher:
+    _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher
+  sweeper:
+    _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper
+    max_batch_size: null
+    params: null
+  help:
+    app_name: ${hydra.job.name}
+    header: '${hydra.help.app_name} is powered by Hydra.
+
+      '
+    footer: 'Powered by Hydra (https://hydra.cc)
+
+      Use --hydra-help to view Hydra specific help
+
+      '
+    template: '${hydra.help.header}
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (group=option)
+
+
+      $APP_CONFIG_GROUPS
+
+
+      == Config ==
+
+      Override anything in the config (foo.bar=value)
+
+
+      $CONFIG
+
+
+      ${hydra.help.footer}
+
+      '
+  hydra_help:
+    template: 'Hydra (${hydra.runtime.version})
+
+      See https://hydra.cc for more info.
+
+
+      == Flags ==
+
+      $FLAGS_HELP
+
+
+      == Configuration groups ==
+
+      Compose your configuration from those groups (For example, append hydra/job_logging=disabled
+      to command line)
+
+
+      $HYDRA_CONFIG_GROUPS
+
+
+      Use ''--cfg hydra'' to Show the Hydra config.
+
+      '
+    hydra_help: ???
+  hydra_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][HYDRA] %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+    root:
+      level: INFO
+      handlers:
+      - console
+    loggers:
+      logging_example:
+        level: DEBUG
+    disable_existing_loggers: false
+  job_logging:
+    version: 1
+    formatters:
+      simple:
+        format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s'
+    handlers:
+      console:
+        class: logging.StreamHandler
+        formatter: simple
+        stream: ext://sys.stdout
+      file:
+        class: logging.FileHandler
+        formatter: simple
+        filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log
+    root:
+      level: INFO
+      handlers:
+      - console
+      - file
+    disable_existing_loggers: false
+  env: {}
+  mode: RUN
+  searchpath: []
+  callbacks: {}
+  output_subdir: .hydra
+  overrides:
+    hydra:
+    - hydra.run.dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra
+    - hydra.mode=RUN
+    task:
+    - mode=sample_eval
+    - seed=42
+    - data=lm1b-wrap
+    - model=small
+    - model.length=128
+    - loader.batch_size=16
+    - loader.eval_batch_size=2
+    - sampling.num_sample_batches=1
+    - sampling.steps=1024
+    - eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt
+    - trainer.devices=1
+    - checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra
+    - +wandb.offline=true
+    - eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_samples.json
+    - algo=candi
+    - algo.sampler=cached
+    - algo.mixed_coeff=0.5
+    - algo.step_size=1.0
+    - algo.temp=1.0
+    - algo.use_percentile_scheduling=True
+    - sampling.noise_removal=ancestral
+    - trainer.accumulate_grad_batches=1
+    - eval.disable_ema=False
+  job:
+    name: dlb_pinned_upstream_main
+    chdir: true
+    override_dirname: +wandb.offline=true,algo.mixed_coeff=0.5,algo.sampler=cached,algo.step_size=1.0,algo.temp=1.0,algo.use_percentile_scheduling=True,algo=candi,checkpointing.save_dir=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra,data=lm1b-wrap,eval.checkpoint_path=/home/ruizzhao/diffusion_baseline/checkpoints/reference_reproduction/flm_baselines/lm1b/lm1b_CANDI.ckpt,eval.disable_ema=False,eval.generated_samples_path=/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_samples.json,loader.batch_size=16,loader.eval_batch_size=2,mode=sample_eval,model.length=128,model=small,sampling.noise_removal=ancestral,sampling.num_sample_batches=1,sampling.steps=1024,seed=42,trainer.accumulate_grad_batches=1,trainer.devices=1
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /home/ruizzhao/diffusion_baseline
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /home/ruizzhao/diffusion_baseline/upstreams/candi/configs
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_1024/upstream_hydra
+    choices:
+      algo: candi
+      prior: none
+      lr_scheduler: constant
+      noise: log-linear
+      strategy: ddp
+      model: small
+      data: lm1b-wrap
+      hydra/env: default
+      hydra/callbacks: null
+      hydra/job_logging: default
+      hydra/hydra_logging: default
+      hydra/hydra_help: default
+      hydra/help: default
+      hydra/sweeper: basic
+      hydra/launcher: basic
+      hydra/output: default
+  verbose: false

results/smoke/samples/lm1b/candi/steps_16/conversion_metadata.json

@@ -0,0 +1,26 @@
+{
+  "checkpoint_lock_id": "flm_lm1b_reproductions:3549a2eff7c0b0ccceca8d69a22baf7795fd9843fa9fcad41a8ddc186b9892f0:lm1b_CANDI.ckpt",
+  "checkpoint_selection": {
+    "path": "lm1b_CANDI.ckpt",
+    "resource": "flm_lm1b_reproductions",
+    "teacher_family": "hybrid_candi"
+  },
+  "checkpoint_sha256": "48eb0fdd4db073b4bd32b194fdaa9216068d4d5b4e7ea5073c790b1a38604598",
+  "exclude_from_latency": true,
+  "format": "upstream-generated-seqs-v1",
+  "generated_samples": 2,
+  "generation_seconds_sentinel": 0.0,
+  "generation_seconds_source": "unavailable_excluded_sentinel",
+  "requested_samples": 1,
+  "teacher_family": "hybrid_candi",
+  "token_ids_source": "upstream",
+  "token_ids_transformation": {
+    "max_length": 128,
+    "operation": "validated_exact_length"
+  },
+  "tokenizer": "bert-base-uncased",
+  "tokenizer_revision": "86b5e0934494bd15c9632b12f734a8a67f723594",
+  "trim_policy": "stable_prefix",
+  "trimmed_samples": 1,
+  "upstream_output": "/home/ruizzhao/diffusion_baseline/results/smoke/samples/lm1b/candi/steps_16/upstream_samples.json"
+}