Training in progress, step 54

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

2025-10-10/23-16-13/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: gpt2
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 256
+  context_len: 224
+  num_train_samples: 100000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: your-org/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-16-13/.hydra/hydra.yaml

@@ -0,0 +1,168 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - model.pretrained=gpt2
+    - hfhub.repo_id=your-org/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=256
+    - dataset.context_len=224
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=224,dataset.seq_len=256,experiment=stage1,hfhub.repo_id=your-org/lorentz-poc-stage1,model.pretrained=gpt2,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-16-13
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-16-13/.hydra/overrides.yaml

@@ -0,0 +1,7 @@
+- experiment=stage1
+- model.pretrained=gpt2
+- hfhub.repo_id=your-org/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=256
+- dataset.context_len=224

2025-10-10/23-16-52/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: gpt2
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 256
+  context_len: 224
+  num_train_samples: 100000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: your-org/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-16-52/.hydra/hydra.yaml

@@ -0,0 +1,168 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - model.pretrained=gpt2
+    - hfhub.repo_id=your-org/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=256
+    - dataset.context_len=224
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=224,dataset.seq_len=256,experiment=stage1,hfhub.repo_id=your-org/lorentz-poc-stage1,model.pretrained=gpt2,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-16-52
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-16-52/.hydra/overrides.yaml

@@ -0,0 +1,7 @@
+- experiment=stage1
+- model.pretrained=gpt2
+- hfhub.repo_id=your-org/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=256
+- dataset.context_len=224

2025-10-10/23-18-13/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 100000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-18-13/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-18-13
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-18-13/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-21-52/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 100000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-21-52/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-21-52
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-21-52/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-23-18/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: jkminder/lorentz-poc-stage1
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 2
+  p_intervene: 0.1
+  use_measurement: false
+  force_text: <intro>I am uncertain.</intro>
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage2
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-23-18/.hydra/hydra.yaml

@@ -0,0 +1,168 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage2
+    - model.pretrained=jkminder/lorentz-poc-stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage2
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage2,hfhub.repo_id=jkminder/lorentz-poc-stage2,model.pretrained=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-23-18
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage2
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-23-18/.hydra/overrides.yaml

@@ -0,0 +1,7 @@
+- experiment=stage2
+- model.pretrained=jkminder/lorentz-poc-stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage2
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-23-47/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-23-47/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-23-47
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-23-47/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-28-17/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-28-17/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-28-17
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-28-17/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-33-34/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-33-34/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-33-34
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-33-34/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-35-10/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-35-10/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-35-10
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-35-10/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-36-21/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-36-21/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-36-21
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-36-21/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-40-03/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-40-03/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-40-03
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-40-03/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-44-29/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000

2025-10-10/23-44-29/.hydra/hydra.yaml

@@ -0,0 +1,167 @@
+hydra:
+  run:
+    dir: outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+  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.mode=RUN
+    task:
+    - experiment=stage1
+    - hfhub.repo_id=jkminder/lorentz-poc-stage1
+    - wandb.project=lorentz-forcing
+    - training.num_train_epochs=1
+    - dataset.seq_len=512
+    - dataset.context_len=512
+  job:
+    name: train
+    chdir: null
+    override_dirname: dataset.context_len=512,dataset.seq_len=512,experiment=stage1,hfhub.repo_id=jkminder/lorentz-poc-stage1,training.num_train_epochs=1,wandb.project=lorentz-forcing
+    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: /mnt/nw/home/j.minder/repositories/lorentz_forcing
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /mnt/nw/home/j.minder/repositories/lorentz_forcing/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /mnt/nw/home/j.minder/repositories/lorentz_forcing/outputs/2025-10-10/23-44-29
+    choices:
+      eval: default
+      hfhub: default
+      wandb: default
+      experiment: stage1
+      training: default
+      dataset: fineweb
+      model: llama32_1B
+      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

2025-10-10/23-44-29/.hydra/overrides.yaml

@@ -0,0 +1,6 @@
+- experiment=stage1
+- hfhub.repo_id=jkminder/lorentz-poc-stage1
+- wandb.project=lorentz-forcing
+- training.num_train_epochs=1
+- dataset.seq_len=512
+- dataset.context_len=512

2025-10-10/23-47-52/.hydra/config.yaml

@@ -0,0 +1,45 @@
+model:
+  pretrained: meta-llama/Llama-3.2-1B
+  resize_embeddings: true
+dataset:
+  name: science-of-finetuning/fineweb-1m-sample
+  config: ''
+  seq_len: 512
+  context_len: 512
+  num_train_samples: 1000
+training:
+  output_dir: outputs
+  num_train_epochs: 1
+  per_device_train_batch_size: 4
+  gradient_accumulation_steps: 1
+  learning_rate: 5.0e-05
+  weight_decay: 0.0
+  warmup_steps: 0
+  save_steps: 500
+  logging_steps: 50
+  seed: 42
+  fp16: false
+  bf16: true
+  calibration_batch_size: 128
+experiment:
+  stage: 1
+  p_intervene: 0.1
+  use_measurement: true
+  force_text: null
+  init_from:
+    hub_repo: null
+    local_ckpt: null
+wandb:
+  project: lorentz-forcing
+  entity: jkminder
+hfhub:
+  push_to_hub: true
+  repo_id: jkminder/lorentz-poc-stage1
+eval:
+  output_path: eval_results.json
+  max_eval_samples: 128
+  seed: 42
+  ci:
+    method: bootstrap
+    level: 0.95
+    bootstrap_iters: 1000