Upload folder using huggingface_hub

.gitattributes

@@ -171,3 +171,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 2026.03.17/12.46.50_train_llm_lowdim_box-close-v2/wandb/run-20260317_124654-kqfw3lgo/run-kqfw3lgo.wandb filter=lfs diff=lfs merge=lfs -text
 2026.03.17/16.13.39_train_rbc_reward_model_box-close-v2/logs.json.txt filter=lfs diff=lfs merge=lfs -text
 2026.03.17/16.13.39_train_rbc_reward_model_box-close-v2/wandb/run-20260317_161342-ydactfyz/run-ydactfyz.wandb filter=lfs diff=lfs merge=lfs -text
+2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/run-x1m9280b.wandb filter=lfs diff=lfs merge=lfs -text

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/.hydra/config.yaml

@@ -0,0 +1,115 @@
+name: train_llm_lowdim
+_target_: llmbc.workspace.train_llm_workspace.TrainLLMWorkspace
+obs_dim: ${task.obs_dim}
+action_dim: ${task.action_dim}
+horizon: 1
+n_obs_steps: 1
+n_action_steps: 1
+task_name: ${task.name}
+exp_name: train llm
+model_name: ${llm.name}
+use_quantization: ${llm.use_quantization}
+lora_config: ${llm.lora_config}
+dataset:
+  test_data_ratio: 0.01
+debug: false
+training:
+  seed: 42
+  per_device_train_batch_size: 32
+  per_device_eval_batch_size: 32
+  gradient_accumulation_steps: 4
+  optim: paged_adamw_32bit
+  num_train_epochs: 10
+  eval_strategy: steps
+  logging_steps: 1
+  warmup_steps: 1000
+  logging_strategy: steps
+  learning_rate: 1.0e-05
+  fp16: false
+  bf16: true
+  tf32: true
+  group_by_length: true
+  report_to: wandb
+  save_steps: 5000
+  eval_steps: 10
+  use_joint_mlp_projector: ${llm.use_joint_mlp_projector}
+  joint_obs_action_mlp_lr: 5.0e-06
+trainer:
+  obs_dim: ${obs_dim}
+  action_dim: ${action_dim}
+  use_joint_mlp_projector: ${llm.use_joint_mlp_projector}
+  max_seq_length: ${llm.max_length}
+  dataset_text_field: text
+  packing: false
+logging:
+  project: llm_module_finetuning
+  resume: true
+  mode: online
+  name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name}
+  tags:
+  - ${name}
+  - ${task_name}
+  - ${exp_name}
+  id: null
+  group: null
+multi_run:
+  run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name}
+  wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name}
+task:
+  name: box-close-v2
+  obs_dim: 9
+  action_dim: 4
+  env_runner:
+    _target_: llmbc.env_runner.metaworld_lowdim_runner.MetaworldLowdimRunner
+    env_name: llf-metaworld-box-close-v2
+    n_train: 10
+    n_test: 50
+    n_envs: 10
+    max_steps: 30
+    n_obs_steps: ${n_obs_steps}
+    n_action_steps: ${n_action_steps}
+    instruction_type: b
+    feedback_type:
+    - hp
+    - hn
+    - fp
+    visual: false
+    discount: 0.9
+  dataset:
+    _target_: llmbc.dataset.metaworld_lowdim_dataset.MetaworldLowdimDataset
+    data_path: datasets/box-close-v2-general.pt
+    data_path2: datasets/box-close-v2.pt
+    horizon: ${horizon}
+    pad_before: ${eval:'${n_obs_steps}-1'}
+    pad_after: ${eval:'${n_action_steps}-1'}
+    obs_eef_target: true
+    use_manual_normalizer: false
+    val_ratio: 0.05
+    dummy_normalizer: false
+  instructor:
+    _target_: llmbc.translator.instructor.metaworld_instructor.box_close_v2_instructor.BoxCloseV2Instructor
+llm:
+  name: mistralai/Mistral-7B-Instruct-v0.3
+  model_name: Mistral-7B-Instruct-v0.3
+  config_target: llmbc.model.llm.mistral_lowdim_model.LowdimMistralConfig
+  causal_lm_target: llmbc.model.llm.mistral_lowdim_model.LowdimMistralForCausalLM
+  use_quantization: true
+  use_joint_mlp_projector: true
+  llm_mode: mlp-finetuned
+  finetune_mode: lora
+  checkpoint: data/outputs/2026.03.14/23.40.55_train_mlp_projector_box-close-v2/checkpoints/latest.ckpt
+  max_length: 100
+  lora_config:
+    r: 8
+    lora_alpha: 16
+    lora_dropout: 0.05
+    bias: none
+    task_type: CAUSAL_LM
+  prompter:
+    _target_: llmbc.translator.prompter.mistral_prompter.MistralPrompter
+    use_joint_mlp_projector: true
+  hydra:
+    job:
+      override_dirname: ${model_name}
+    run:
+      dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/.hydra/hydra.yaml

@@ -0,0 +1,154 @@
+hydra:
+  run:
+    dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name}
+  sweep:
+    dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name}
+    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: []
+  job:
+    name: train
+    chdir: null
+    override_dirname: ''
+    id: ???
+    num: ???
+    config_name: llmdp_llm_box-close-v2_mistral-7b-instruct-v0.3.yaml
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.2.0
+    version_base: '1.2'
+    cwd: /tmp2/chyang/workspace/LLM-BC
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /tmp2/chyang/workspace/LLM-BC/config/llm_backbone
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2
+    choices:
+      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

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/.hydra/overrides.yaml

@@ -0,0 +1 @@
+[]

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/README.md

@@ -0,0 +1,202 @@
+---
+base_model: mistralai/Mistral-7B-Instruct-v0.3
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.14.0

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
+  "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "lm_head",
+    "up_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75753f1f0c86677fa1e6c69e014448ccb5af5655d115db5ac002e6bace0c9d71
+size 154365949

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/README.md

@@ -0,0 +1,202 @@
+---
+base_model: mistralai/Mistral-7B-Instruct-v0.3
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.14.0

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
+  "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "lm_head",
+    "up_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1d3ccbc2d1bff91f403bf2f83d891e0947b255a9729ed4dfa7388db6da427bf
+size 154365949

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/config.json

@@ -0,0 +1,48 @@
+{
+  "_attn_implementation_autoset": true,
+  "_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
+  "action_dim": 4,
+  "architectures": [
+    "MistralForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 32768,
+  "model_type": "mistral_lowdim",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "obs_dim": 9,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": [
+      "joint_obs_action_projector"
+    ],
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.47.1",
+  "use_cache": false,
+  "use_joint_mlp_projector": true,
+  "vocab_size": 32768
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/mlp_projector.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2291fb207b9e8ff0aaab49e12bb914eecd0f0b0f11d5e31ec10dfeeaef263f63
+size 67356864

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a17a019b5ad6b1540bb501e9647ba8fac1045e17e44f5a9a00e8fa4f47d05e3
+size 305112506

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b066b3a37fb18b36ee0d4b205258257d4bcd65310f4c486b427394e2080d99f1
+size 14244

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c50c3e97c48d246a11385d766ea14c9a26fa8602590afacafa00be824daaf5dc
+size 1064

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37f00374dea48658ee8f5d0f21895b9bc55cb0103939607c8185bfd1c6ca1f89
+size 587404

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f633cbc73694f0c37e178ad51e116e8cde4390accf0f74458b31e4225377613b
+size 5944

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/README.md

@@ -0,0 +1,202 @@
+---
+base_model: mistralai/Mistral-7B-Instruct-v0.3
+library_name: peft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.14.0

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/adapter_config.json

@@ -0,0 +1,38 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
+  "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "v_proj",
+    "k_proj",
+    "lm_head",
+    "up_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:75753f1f0c86677fa1e6c69e014448ccb5af5655d115db5ac002e6bace0c9d71
+size 154365949

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/config.json

@@ -0,0 +1,48 @@
+{
+  "_attn_implementation_autoset": true,
+  "_name_or_path": "mistralai/Mistral-7B-Instruct-v0.3",
+  "action_dim": 4,
+  "architectures": [
+    "MistralForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 32768,
+  "model_type": "mistral_lowdim",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "obs_dim": 9,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": true,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": [
+      "joint_obs_action_projector"
+    ],
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.47.1",
+  "use_cache": false,
+  "use_joint_mlp_projector": true,
+  "vocab_size": 32768
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/mlp_projector.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d1a4f49ae341a396e725dc8e7e49e947e282b85ce828ef874ad875c9c7484c1
+size 67356864

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7e2c3c0158d370c6bf9b2a6c447ef9c35bad3f90bee2ac52476d24786e8a693
+size 305112506

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2a94c1a4456a30398e5ea527622f8e055b6818c0582b4ca87eb9e398f200712
+size 14244

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a5e7ff28843a09e39190ed8d73ee956ee41e68c89bd28c0be2d3ceff5c00ef6
+size 1064

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37f00374dea48658ee8f5d0f21895b9bc55cb0103939607c8185bfd1c6ca1f89
+size 587404

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/checkpoint-5880/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f633cbc73694f0c37e178ad51e116e8cde4390accf0f74458b31e4225377613b
+size 5944

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/normalizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a1cb28cf65a776a736481b2a05aa3c4cf0d796a0ec84eaa0b9e4aa06bdc79aa
+size 4514

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/train.log

@@ -0,0 +1,11 @@
+[2026-03-18 22:30:56,869][numexpr.utils][INFO] - Note: NumExpr detected 24 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
+[2026-03-18 22:30:56,869][numexpr.utils][INFO] - NumExpr defaulting to 8 threads.
+[2026-03-18 22:31:00,129][datasets][INFO] - PyTorch version 2.2.2 available.
+[2026-03-18 22:31:00,129][datasets][INFO] - TensorFlow version 2.15.1 available.
+[2026-03-18 22:31:00,130][datasets][INFO] - JAX version 0.4.30 available.
+[2026-03-18 22:31:24,560][datasets.arrow_dataset][WARNING] - Setting TOKENIZERS_PARALLELISM=false for forked processes.
+[2026-03-18 22:31:35,877][datasets.arrow_dataset][WARNING] - Setting TOKENIZERS_PARALLELISM=false for forked processes.
+[2026-03-18 22:31:41,531][root][INFO] - gcc -pthread -B /home/chyang/miniconda3/envs/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /home/chyang/miniconda3/envs/llm-bc/include -I/home/chyang/miniconda3/envs/llm-bc/include -fPIC -O2 -isystem /home/chyang/miniconda3/envs/llm-bc/include -fPIC -c /tmp/tmph6bq3krg/test.c -o /tmp/tmph6bq3krg/test.o
+[2026-03-18 22:31:41,627][root][INFO] - gcc -pthread -B /home/chyang/miniconda3/envs/llm-bc/compiler_compat /tmp/tmph6bq3krg/test.o -laio -o /tmp/tmph6bq3krg/a.out
+[2026-03-18 22:31:42,479][root][INFO] - gcc -pthread -B /home/chyang/miniconda3/envs/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /home/chyang/miniconda3/envs/llm-bc/include -I/home/chyang/miniconda3/envs/llm-bc/include -fPIC -O2 -isystem /home/chyang/miniconda3/envs/llm-bc/include -fPIC -c /tmp/tmpqkrxt68a/test.c -o /tmp/tmpqkrxt68a/test.o
+[2026-03-18 22:31:42,552][root][INFO] - gcc -pthread -B /home/chyang/miniconda3/envs/llm-bc/compiler_compat /tmp/tmpqkrxt68a/test.o -L/usr/local/cuda -L/usr/local/cuda/lib64 -lcufile -o /tmp/tmpqkrxt68a/a.out

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/debug-internal.log

@@ -0,0 +1,19 @@
+{"time":"2026-03-18T22:31:01.204991579+08:00","level":"INFO","msg":"using version","core version":"0.18.6"}
+{"time":"2026-03-18T22:31:01.205010756+08:00","level":"INFO","msg":"created symlink","path":"/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-core.log"}
+{"time":"2026-03-18T22:31:01.312346847+08:00","level":"INFO","msg":"created new stream","id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312367301+08:00","level":"INFO","msg":"stream: started","id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312665109+08:00","level":"INFO","msg":"sender: started","stream_id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312430257+08:00","level":"INFO","msg":"handler: started","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-18T22:31:01.312656373+08:00","level":"INFO","msg":"writer: Do: started","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-18T22:31:02.758068808+08:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2026-03-19T04:22:27.963176902+08:00","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b/file_stream"}
+{"time":"2026-03-19T05:59:47.962016684+08:00","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b/file_stream"}
+{"time":"2026-03-19T17:31:14.776552794+08:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2026-03-19T17:31:14.777008363+08:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2026-03-19T17:31:15.776985704+08:00","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"uploading wandb-summary.json","runtime_seconds":0.192680949,"progress":"515B/515B"},{"desc":"saving job artifact","runtime_seconds":0.014816469}],"total_operations":2}}
+{"time":"2026-03-19T17:31:21.307731838+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-19T17:31:24.586012627+08:00","level":"INFO","msg":"stream: closing","id":"x1m9280b"}
+{"time":"2026-03-19T17:31:24.586046207+08:00","level":"INFO","msg":"handler: closed","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-19T17:31:24.586077373+08:00","level":"INFO","msg":"writer: Close: closed","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-19T17:31:24.586176871+08:00","level":"INFO","msg":"sender: closed","stream_id":"x1m9280b"}
+{"time":"2026-03-19T17:31:24.586233462+08:00","level":"INFO","msg":"stream: closed","id":"x1m9280b"}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/debug.log

@@ -0,0 +1,35 @@
+2026-03-18 22:31:01,198 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Current SDK version is 0.18.6
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Configure stats pid to 2041405
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from /home/chyang/.config/wandb/settings
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from /tmp2/chyang/workspace/LLM-BC/wandb/settings
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from environment variables: {}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Applying setup settings: {'mode': 'online', '_disable_service': None}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Inferring run settings from compute environment: {'program_relpath': 'train.py', 'program_abspath': '/tmp2/chyang/workspace/LLM-BC/train.py', 'program': '/tmp2/chyang/workspace/LLM-BC/./train.py'}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Applying login settings: {}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:_log_setup():533] Logging user logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug.log
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:_log_setup():534] Logging internal logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-internal.log
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():619] calling init triggers
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():626] wandb.init called with sweep_config: {}
+config: {'name': 'train_llm_lowdim', '_target_': 'llmbc.workspace.train_llm_workspace.TrainLLMWorkspace', 'obs_dim': 9, 'action_dim': 4, 'horizon': 1, 'n_obs_steps': 1, 'n_action_steps': 1, 'task_name': 'box-close-v2', 'exp_name': 'train llm', 'model_name': 'mistralai/Mistral-7B-Instruct-v0.3', 'use_quantization': True, 'lora_config': {'r': 8, 'lora_alpha': 16, 'lora_dropout': 0.05, 'bias': 'none', 'task_type': 'CAUSAL_LM'}, 'dataset': {'test_data_ratio': 0.01}, 'debug': False, 'training': {'seed': 42, 'per_device_train_batch_size': 32, 'per_device_eval_batch_size': 32, 'gradient_accumulation_steps': 4, 'optim': 'paged_adamw_32bit', 'num_train_epochs': 10, 'eval_strategy': 'steps', 'logging_steps': 1, 'warmup_steps': 1000, 'logging_strategy': 'steps', 'learning_rate': 1e-05, 'fp16': False, 'bf16': True, 'tf32': True, 'group_by_length': True, 'report_to': 'wandb', 'save_steps': 5000, 'eval_steps': 10, 'use_joint_mlp_projector': True, 'joint_obs_action_mlp_lr': 5e-06}, 'trainer': {'obs_dim': 9, 'action_dim': 4, 'use_joint_mlp_projector': True, 'max_seq_length': 100, 'dataset_text_field': 'text', 'packing': False}, 'logging': {'project': 'llm_module_finetuning', 'resume': True, 'mode': 'online', 'name': '2026.03.18-22.30.56_train_llm_lowdim_box-close-v2', 'tags': ['train_llm_lowdim', 'box-close-v2', 'train llm'], 'id': None, 'group': None}, 'multi_run': {'run_dir': 'data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2', 'wandb_name_base': '2026.03.18-22.30.56_train_llm_lowdim_box-close-v2'}, 'task': {'name': 'box-close-v2', 'obs_dim': 9, 'action_dim': 4, 'env_runner': {'_target_': 'llmbc.env_runner.metaworld_lowdim_runner.MetaworldLowdimRunner', 'env_name': 'llf-metaworld-box-close-v2', 'n_train': 10, 'n_test': 50, 'n_envs': 10, 'max_steps': 30, 'n_obs_steps': 1, 'n_action_steps': 1, 'instruction_type': 'b', 'feedback_type': ['hp', 'hn', 'fp'], 'visual': False, 'discount': 0.9}, 'dataset': {'_target_': 'llmbc.dataset.metaworld_lowdim_dataset.MetaworldLowdimDataset', 'data_path': 'datasets/box-close-v2-general.pt', 'data_path2': 'datasets/box-close-v2.pt', 'horizon': 1, 'pad_before': 0, 'pad_after': 0, 'obs_eef_target': True, 'use_manual_normalizer': False, 'val_ratio': 0.05, 'dummy_normalizer': False}, 'instructor': {'_target_': 'llmbc.translator.instructor.metaworld_instructor.box_close_v2_instructor.BoxCloseV2Instructor'}}, 'llm': {'name': 'mistralai/Mistral-7B-Instruct-v0.3', 'model_name': 'Mistral-7B-Instruct-v0.3', 'config_target': 'llmbc.model.llm.mistral_lowdim_model.LowdimMistralConfig', 'causal_lm_target': 'llmbc.model.llm.mistral_lowdim_model.LowdimMistralForCausalLM', 'use_quantization': True, 'use_joint_mlp_projector': True, 'llm_mode': 'mlp-finetuned', 'finetune_mode': 'lora', 'checkpoint': 'data/outputs/2026.03.14/23.40.55_train_mlp_projector_box-close-v2/checkpoints/latest.ckpt', 'max_length': 100, 'lora_config': {'r': 8, 'lora_alpha': 16, 'lora_dropout': 0.05, 'bias': 'none', 'task_type': 'CAUSAL_LM'}, 'prompter': {'_target_': 'llmbc.translator.prompter.mistral_prompter.MistralPrompter', 'use_joint_mlp_projector': True}, 'hydra': {'job': {'override_dirname': 'mistralai/Mistral-7B-Instruct-v0.3'}, 'run': {'dir': 'data/outputs/2026.03.18/22.30.56_mistralai/Mistral-7B-Instruct-v0.3'}}}}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():669] starting backend
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():673] sending inform_init request
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [backend.py:_multiprocessing_setup():104] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2026-03-18 22:31:01,200 INFO    MainThread:2041405 [wandb_init.py:init():686] backend started and connected
+2026-03-18 22:31:01,204 INFO    MainThread:2041405 [wandb_init.py:init():781] updated telemetry
+2026-03-18 22:31:01,224 INFO    MainThread:2041405 [wandb_init.py:init():814] communicating run to backend with 90.0 second timeout
+2026-03-18 22:31:02,754 INFO    MainThread:2041405 [wandb_init.py:init():867] starting run threads in backend
+2026-03-18 22:31:02,923 INFO    MainThread:2041405 [wandb_run.py:_console_start():2451] atexit reg
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2299] redirect: wrap_raw
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2364] Wrapping output streams.
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2389] Redirects installed.
+2026-03-18 22:31:02,925 INFO    MainThread:2041405 [wandb_init.py:init():911] run started, returning control to user process
+2026-03-18 22:31:51,682 INFO    MainThread:2041405 [wandb_run.py:_config_callback():1389] config_cb None None {'peft_config': {'default': {'task_type': 'CAUSAL_LM', 'peft_type': <PeftType.LORA: 'LORA'>, 'auto_mapping': None, 'base_model_name_or_path': 'mistralai/Mistral-7B-Instruct-v0.3', 'revision': None, 'inference_mode': False, 'r': 8, 'target_modules': {'down_proj', 'v_proj', 'k_proj', 'lm_head', 'up_proj', 'q_proj', 'gate_proj', 'o_proj'}, 'exclude_modules': None, 'lora_alpha': 16, 'lora_dropout': 0.05, 'fan_in_fan_out': False, 'bias': 'none', 'use_rslora': False, 'modules_to_save': None, 'init_lora_weights': True, 'layers_to_transform': None, 'layers_pattern': None, 'rank_pattern': {}, 'alpha_pattern': {}, 'megatron_config': None, 'megatron_core': 'megatron.core', 'loftq_config': {}, 'eva_config': None, 'use_dora': False, 'layer_replication': None, 'runtime_config': {'ephemeral_gpu_offload': False}, 'lora_bias': False}}, 'obs_dim': 9, 'action_dim': 4, 'use_joint_mlp_projector': True, 'vocab_size': 32768, 'max_position_embeddings': 32768, 'hidden_size': 4096, 'intermediate_size': 14336, 'num_hidden_layers': 32, 'num_attention_heads': 32, 'sliding_window': None, 'head_dim': 128, 'num_key_value_heads': 8, 'hidden_act': 'silu', 'initializer_range': 0.02, 'rms_norm_eps': 1e-05, 'use_cache': False, 'rope_theta': 1000000.0, 'attention_dropout': 0.0, 'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'bfloat16', 'use_bfloat16': False, 'tf_legacy_loss': False, 'pruned_heads': {}, 'tie_word_embeddings': False, 'chunk_size_feed_forward': 0, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'suppress_tokens': None, 'begin_suppress_tokens': None, 'architectures': ['MistralForCausalLM'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': None, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'mistralai/Mistral-7B-Instruct-v0.3', '_attn_implementation_autoset': True, 'transformers_version': '4.47.1', 'model_type': 'mistral_lowdim', 'quantization_config': {'quant_method': 'BITS_AND_BYTES', '_load_in_8bit': False, '_load_in_4bit': True, 'llm_int8_threshold': 6.0, 'llm_int8_skip_modules': ['joint_obs_action_projector'], 'llm_int8_enable_fp32_cpu_offload': False, 'llm_int8_has_fp16_weight': False, 'bnb_4bit_quant_type': 'nf4', 'bnb_4bit_use_double_quant': True, 'bnb_4bit_compute_dtype': 'bfloat16', 'bnb_4bit_quant_storage': 'uint8', 'load_in_4bit': True, 'load_in_8bit': False}, 'output_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': True, 'do_predict': False, 'eval_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 32, 'per_device_eval_batch_size': 32, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 4, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 10, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.0, 'warmup_steps': 1000, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/runs/Mar18_22-31-41_A6000-2', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 1, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 5000, 'save_total_limit': None, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': True, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': 10, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}, 'deepspeed': None, 'label_smoothing_factor': 0.0, 'optim': 'paged_adamw_32bit', 'optim_args': None, 'adafactor': False, 'group_by_length': True, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': None, 'hub_model_id': None, 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': False, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'evaluation_strategy': None, 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'dispatch_batches': None, 'split_batches': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules': None, 'batch_eval_metrics': False, 'eval_on_start': False, 'use_liger_kernel': False, 'eval_use_gather_object': False, 'average_tokens_across_devices': False, 'dataset_text_field': 'text', 'packing': False, 'max_seq_length': 100, 'dataset_num_proc': None, 'dataset_batch_size': 1000, 'model_init_kwargs': None, 'dataset_kwargs': {}, 'eval_packing': None, 'num_of_sequences': 1024, 'chars_per_token': '<CHARS_PER_TOKEN>', 'use_liger': False, 'joint_obs_action_mlp_lr': 5e-06, 'obs_mlp_lr': None, 'action_mlp_lr': None}
+2026-03-18 22:31:51,686 INFO    MainThread:2041405 [wandb_config.py:__setitem__():154] config set model/num_parameters = 7286128640 - <bound method Run._config_callback of <wandb.sdk.wandb_run.Run object at 0x7fe2561236a0>>
+2026-03-18 22:31:51,686 INFO    MainThread:2041405 [wandb_run.py:_config_callback():1389] config_cb model/num_parameters 7286128640 None
+2026-03-19 17:31:14,775 INFO    MainThread:2041405 [wandb_run.py:_finish():2146] finishing run chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_atexit_cleanup():2414] got exitcode: 0
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_restore():2396] restore
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_restore():2402] restore done
+2026-03-19 17:31:24,580 INFO    MainThread:2041405 [wandb_run.py:_footer_history_summary_info():3963] rendering history
+2026-03-19 17:31:24,581 INFO    MainThread:2041405 [wandb_run.py:_footer_history_summary_info():3995] rendering summary
+2026-03-19 17:31:24,585 INFO    MainThread:2041405 [wandb_run.py:_footer_sync_info():3922] logging synced files

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/files/config.yaml

@@ -0,0 +1,749 @@
+_attn_implementation_autoset:
+    value: true
+_name_or_path:
+    value: mistralai/Mistral-7B-Instruct-v0.3
+_target_:
+    value: llmbc.workspace.train_llm_workspace.TrainLLMWorkspace
+_wandb:
+    value:
+        cli_version: 0.18.6
+        m:
+            - "1": eval/runtime
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/global_step
+              "6":
+                - 3
+              "7": []
+            - "1": eval/loss
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": eval/steps_per_second
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/grad_norm
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/loss
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/learning_rate
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": train/epoch
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+            - "1": eval/samples_per_second
+              "5": 2
+              "6":
+                - 1
+                - 3
+              "7": []
+        python_version: 3.9.20
+        t:
+            "1":
+                - 1
+                - 2
+                - 3
+                - 5
+                - 11
+                - 12
+                - 41
+                - 49
+                - 50
+                - 51
+                - 53
+                - 55
+                - 71
+                - 84
+                - 98
+            "2":
+                - 1
+                - 2
+                - 3
+                - 5
+                - 11
+                - 12
+                - 41
+                - 49
+                - 50
+                - 51
+                - 53
+                - 55
+                - 71
+                - 84
+                - 98
+            "3":
+                - 2
+                - 7
+                - 13
+                - 15
+                - 16
+                - 19
+                - 23
+                - 55
+                - 62
+                - 66
+            "4": 3.9.20
+            "5": 0.18.6
+            "6": 4.47.1
+            "8":
+                - 5
+            "9":
+                "1": transformers_trainer
+            "12": 0.18.6
+            "13": linux-x86_64
+accelerator_config:
+    value:
+        dispatch_batches: null
+        even_batches: true
+        gradient_accumulation_kwargs: null
+        non_blocking: false
+        split_batches: false
+        use_seedable_sampler: true
+action_dim:
+    value: 4
+action_mlp_lr:
+    value: null
+adafactor:
+    value: false
+adam_beta1:
+    value: 0.9
+adam_beta2:
+    value: 0.999
+adam_epsilon:
+    value: 1e-08
+add_cross_attention:
+    value: false
+architectures:
+    value:
+        - MistralForCausalLM
+attention_dropout:
+    value: 0
+auto_find_batch_size:
+    value: false
+average_tokens_across_devices:
+    value: false
+bad_words_ids:
+    value: null
+batch_eval_metrics:
+    value: false
+begin_suppress_tokens:
+    value: null
+bf16:
+    value: true
+bf16_full_eval:
+    value: false
+bos_token_id:
+    value: 1
+chars_per_token:
+    value: <CHARS_PER_TOKEN>
+chunk_size_feed_forward:
+    value: 0
+cross_attention_hidden_size:
+    value: null
+data_seed:
+    value: null
+dataloader_drop_last:
+    value: false
+dataloader_num_workers:
+    value: 0
+dataloader_persistent_workers:
+    value: false
+dataloader_pin_memory:
+    value: true
+dataloader_prefetch_factor:
+    value: null
+dataset:
+    value:
+        test_data_ratio: 0.01
+dataset_batch_size:
+    value: 1000
+dataset_num_proc:
+    value: null
+dataset_text_field:
+    value: text
+ddp_backend:
+    value: null
+ddp_broadcast_buffers:
+    value: null
+ddp_bucket_cap_mb:
+    value: null
+ddp_find_unused_parameters:
+    value: null
+ddp_timeout:
+    value: 1800
+debug:
+    value: []
+decoder_start_token_id:
+    value: null
+deepspeed:
+    value: null
+disable_tqdm:
+    value: false
+dispatch_batches:
+    value: null
+diversity_penalty:
+    value: 0
+do_eval:
+    value: true
+do_predict:
+    value: false
+do_sample:
+    value: false
+do_train:
+    value: false
+early_stopping:
+    value: false
+encoder_no_repeat_ngram_size:
+    value: 0
+eos_token_id:
+    value: 2
+eval_accumulation_steps:
+    value: null
+eval_delay:
+    value: 0
+eval_do_concat_batches:
+    value: true
+eval_on_start:
+    value: false
+eval_packing:
+    value: null
+eval_steps:
+    value: 10
+eval_strategy:
+    value: steps
+eval_use_gather_object:
+    value: false
+evaluation_strategy:
+    value: null
+exp_name:
+    value: train llm
+exponential_decay_length_penalty:
+    value: null
+finetuning_task:
+    value: null
+forced_bos_token_id:
+    value: null
+forced_eos_token_id:
+    value: null
+fp16:
+    value: false
+fp16_backend:
+    value: auto
+fp16_full_eval:
+    value: false
+fp16_opt_level:
+    value: O1
+fsdp:
+    value: []
+fsdp_config:
+    value:
+        min_num_params: 0
+        xla: false
+        xla_fsdp_grad_ckpt: false
+        xla_fsdp_v2: false
+fsdp_min_num_params:
+    value: 0
+fsdp_transformer_layer_cls_to_wrap:
+    value: null
+full_determinism:
+    value: false
+gradient_accumulation_steps:
+    value: 4
+gradient_checkpointing:
+    value: false
+gradient_checkpointing_kwargs:
+    value: null
+greater_is_better:
+    value: null
+group_by_length:
+    value: true
+half_precision_backend:
+    value: auto
+head_dim:
+    value: 128
+hidden_act:
+    value: silu
+hidden_size:
+    value: 4096
+horizon:
+    value: 1
+hub_always_push:
+    value: false
+hub_model_id:
+    value: null
+hub_private_repo:
+    value: null
+hub_strategy:
+    value: every_save
+hub_token:
+    value: <HUB_TOKEN>
+id2label:
+    value:
+        "0": LABEL_0
+        "1": LABEL_1
+ignore_data_skip:
+    value: false
+include_for_metrics:
+    value: []
+include_inputs_for_metrics:
+    value: false
+include_num_input_tokens_seen:
+    value: false
+include_tokens_per_second:
+    value: false
+initializer_range:
+    value: 0.02
+intermediate_size:
+    value: 14336
+is_decoder:
+    value: false
+is_encoder_decoder:
+    value: false
+jit_mode_eval:
+    value: false
+joint_obs_action_mlp_lr:
+    value: 5e-06
+label_names:
+    value: null
+label_smoothing_factor:
+    value: 0
+label2id:
+    value:
+        LABEL_0: 0
+        LABEL_1: 1
+learning_rate:
+    value: 1e-05
+length_column_name:
+    value: length
+length_penalty:
+    value: 1
+llm:
+    value:
+        causal_lm_target: llmbc.model.llm.mistral_lowdim_model.LowdimMistralForCausalLM
+        checkpoint: data/outputs/2026.03.14/23.40.55_train_mlp_projector_box-close-v2/checkpoints/latest.ckpt
+        config_target: llmbc.model.llm.mistral_lowdim_model.LowdimMistralConfig
+        finetune_mode: lora
+        hydra:
+            job:
+                override_dirname: mistralai/Mistral-7B-Instruct-v0.3
+            run:
+                dir: data/outputs/2026.03.18/22.30.56_mistralai/Mistral-7B-Instruct-v0.3
+        llm_mode: mlp-finetuned
+        lora_config:
+            bias: none
+            lora_alpha: 16
+            lora_dropout: 0.05
+            r: 8
+            task_type: CAUSAL_LM
+        max_length: 100
+        model_name: Mistral-7B-Instruct-v0.3
+        name: mistralai/Mistral-7B-Instruct-v0.3
+        prompter:
+            _target_: llmbc.translator.prompter.mistral_prompter.MistralPrompter
+            use_joint_mlp_projector: true
+        use_joint_mlp_projector: true
+        use_quantization: true
+load_best_model_at_end:
+    value: false
+local_rank:
+    value: 0
+log_level:
+    value: passive
+log_level_replica:
+    value: warning
+log_on_each_node:
+    value: true
+logging:
+    value:
+        group: null
+        id: null
+        mode: online
+        name: 2026.03.18-22.30.56_train_llm_lowdim_box-close-v2
+        project: llm_module_finetuning
+        resume: true
+        tags:
+            - train_llm_lowdim
+            - box-close-v2
+            - train llm
+logging_dir:
+    value: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/runs/Mar18_22-31-41_A6000-2
+logging_first_step:
+    value: false
+logging_nan_inf_filter:
+    value: true
+logging_steps:
+    value: 1
+logging_strategy:
+    value: steps
+lora_config:
+    value:
+        bias: none
+        lora_alpha: 16
+        lora_dropout: 0.05
+        r: 8
+        task_type: CAUSAL_LM
+lr_scheduler_type:
+    value: linear
+max_grad_norm:
+    value: 1
+max_length:
+    value: 20
+max_position_embeddings:
+    value: 32768
+max_seq_length:
+    value: 100
+max_steps:
+    value: -1
+metric_for_best_model:
+    value: null
+min_length:
+    value: 0
+model/num_parameters:
+    value: 7286128640
+model_init_kwargs:
+    value: null
+model_name:
+    value: mistralai/Mistral-7B-Instruct-v0.3
+model_type:
+    value: mistral_lowdim
+mp_parameters:
+    value: ""
+multi_run:
+    value:
+        run_dir: data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2
+        wandb_name_base: 2026.03.18-22.30.56_train_llm_lowdim_box-close-v2
+n_action_steps:
+    value: 1
+n_obs_steps:
+    value: 1
+name:
+    value: train_llm_lowdim
+neftune_noise_alpha:
+    value: null
+no_cuda:
+    value: false
+no_repeat_ngram_size:
+    value: 0
+num_attention_heads:
+    value: 32
+num_beam_groups:
+    value: 1
+num_beams:
+    value: 1
+num_hidden_layers:
+    value: 32
+num_key_value_heads:
+    value: 8
+num_of_sequences:
+    value: 1024
+num_return_sequences:
+    value: 1
+num_train_epochs:
+    value: 10
+obs_dim:
+    value: 9
+obs_mlp_lr:
+    value: null
+optim:
+    value: paged_adamw_32bit
+optim_args:
+    value: null
+optim_target_modules:
+    value: null
+output_attentions:
+    value: false
+output_dir:
+    value: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2
+output_hidden_states:
+    value: false
+output_scores:
+    value: false
+overwrite_output_dir:
+    value: false
+packing:
+    value: false
+pad_token_id:
+    value: null
+past_index:
+    value: -1
+peft_config:
+    value:
+        default:
+            auto_mapping: null
+            base_model_name_or_path: mistralai/Mistral-7B-Instruct-v0.3
+            bias: none
+            eva_config: null
+            exclude_modules: null
+            fan_in_fan_out: false
+            inference_mode: false
+            init_lora_weights: true
+            layer_replication: null
+            layers_pattern: null
+            layers_to_transform: null
+            lora_alpha: 16
+            lora_bias: false
+            lora_dropout: 0.05
+            megatron_config: null
+            megatron_core: megatron.core
+            modules_to_save: null
+            peft_type: LORA
+            r: 8
+            revision: null
+            runtime_config:
+                ephemeral_gpu_offload: false
+            target_modules:
+                - down_proj
+                - v_proj
+                - k_proj
+                - lm_head
+                - up_proj
+                - q_proj
+                - gate_proj
+                - o_proj
+            task_type: CAUSAL_LM
+            use_dora: false
+            use_rslora: false
+per_device_eval_batch_size:
+    value: 32
+per_device_train_batch_size:
+    value: 32
+per_gpu_eval_batch_size:
+    value: null
+per_gpu_train_batch_size:
+    value: null
+prediction_loss_only:
+    value: false
+prefix:
+    value: null
+problem_type:
+    value: null
+push_to_hub:
+    value: false
+push_to_hub_model_id:
+    value: null
+push_to_hub_organization:
+    value: null
+push_to_hub_token:
+    value: <PUSH_TO_HUB_TOKEN>
+quantization_config:
+    value:
+        _load_in_4bit: true
+        _load_in_8bit: false
+        bnb_4bit_compute_dtype: bfloat16
+        bnb_4bit_quant_storage: uint8
+        bnb_4bit_quant_type: nf4
+        bnb_4bit_use_double_quant: true
+        llm_int8_enable_fp32_cpu_offload: false
+        llm_int8_has_fp16_weight: false
+        llm_int8_skip_modules:
+            - joint_obs_action_projector
+        llm_int8_threshold: 6
+        load_in_4bit: true
+        load_in_8bit: false
+        quant_method: BITS_AND_BYTES
+ray_scope:
+    value: last
+remove_invalid_values:
+    value: false
+remove_unused_columns:
+    value: true
+repetition_penalty:
+    value: 1
+report_to:
+    value:
+        - wandb
+restore_callback_states_from_checkpoint:
+    value: false
+resume_from_checkpoint:
+    value: null
+return_dict:
+    value: true
+return_dict_in_generate:
+    value: false
+rms_norm_eps:
+    value: 1e-05
+rope_theta:
+    value: 1e+06
+run_name:
+    value: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2
+save_on_each_node:
+    value: false
+save_only_model:
+    value: false
+save_safetensors:
+    value: true
+save_steps:
+    value: 5000
+save_strategy:
+    value: steps
+save_total_limit:
+    value: null
+seed:
+    value: 42
+sep_token_id:
+    value: null
+skip_memory_metrics:
+    value: true
+sliding_window:
+    value: null
+split_batches:
+    value: null
+suppress_tokens:
+    value: null
+task:
+    value:
+        action_dim: 4
+        dataset:
+            _target_: llmbc.dataset.metaworld_lowdim_dataset.MetaworldLowdimDataset
+            data_path: datasets/box-close-v2-general.pt
+            data_path2: datasets/box-close-v2.pt
+            dummy_normalizer: false
+            horizon: 1
+            obs_eef_target: true
+            pad_after: 0
+            pad_before: 0
+            use_manual_normalizer: false
+            val_ratio: 0.05
+        env_runner:
+            _target_: llmbc.env_runner.metaworld_lowdim_runner.MetaworldLowdimRunner
+            discount: 0.9
+            env_name: llf-metaworld-box-close-v2
+            feedback_type:
+                - hp
+                - hn
+                - fp
+            instruction_type: b
+            max_steps: 30
+            n_action_steps: 1
+            n_envs: 10
+            n_obs_steps: 1
+            n_test: 50
+            n_train: 10
+            visual: false
+        instructor:
+            _target_: llmbc.translator.instructor.metaworld_instructor.box_close_v2_instructor.BoxCloseV2Instructor
+        name: box-close-v2
+        obs_dim: 9
+task_name:
+    value: box-close-v2
+task_specific_params:
+    value: null
+temperature:
+    value: 1
+tf_legacy_loss:
+    value: false
+tf32:
+    value: true
+tie_encoder_decoder:
+    value: false
+tie_word_embeddings:
+    value: false
+tokenizer_class:
+    value: null
+top_k:
+    value: 50
+top_p:
+    value: 1
+torch_compile:
+    value: false
+torch_compile_backend:
+    value: null
+torch_compile_mode:
+    value: null
+torch_dtype:
+    value: bfloat16
+torch_empty_cache_steps:
+    value: null
+torchdynamo:
+    value: null
+torchscript:
+    value: false
+tpu_metrics_debug:
+    value: false
+tpu_num_cores:
+    value: null
+trainer:
+    value:
+        action_dim: 4
+        dataset_text_field: text
+        max_seq_length: 100
+        obs_dim: 9
+        packing: false
+        use_joint_mlp_projector: true
+training:
+    value:
+        bf16: true
+        eval_steps: 10
+        eval_strategy: steps
+        fp16: false
+        gradient_accumulation_steps: 4
+        group_by_length: true
+        joint_obs_action_mlp_lr: 5e-06
+        learning_rate: 1e-05
+        logging_steps: 1
+        logging_strategy: steps
+        num_train_epochs: 10
+        optim: paged_adamw_32bit
+        per_device_eval_batch_size: 32
+        per_device_train_batch_size: 32
+        report_to: wandb
+        save_steps: 5000
+        seed: 42
+        tf32: true
+        use_joint_mlp_projector: true
+        warmup_steps: 1000
+transformers_version:
+    value: 4.47.1
+typical_p:
+    value: 1
+use_bfloat16:
+    value: false
+use_cache:
+    value: false
+use_cpu:
+    value: false
+use_ipex:
+    value: false
+use_joint_mlp_projector:
+    value: true
+use_legacy_prediction_loop:
+    value: false
+use_liger:
+    value: false
+use_liger_kernel:
+    value: false
+use_mps_device:
+    value: false
+use_quantization:
+    value: true
+vocab_size:
+    value: 32768
+warmup_ratio:
+    value: 0
+warmup_steps:
+    value: 1000
+weight_decay:
+    value: 0

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/files/wandb-metadata.json

@@ -0,0 +1,55 @@
+{
+  "os": "Linux-6.8.0-101-generic-x86_64-with-glibc2.35",
+  "python": "3.9.20",
+  "startedAt": "2026-03-18T14:31:01.200411Z",
+  "args": [
+    "--config-path",
+    "config/llm_backbone",
+    "--config-name",
+    "llmdp_llm_box-close-v2_mistral-7b-instruct-v0.3.yaml"
+  ],
+  "program": "/tmp2/chyang/workspace/LLM-BC/./train.py",
+  "codePath": "train.py",
+  "git": {
+    "remote": "https://github.com/CHYang25/LLM-BC.git",
+    "commit": "2fc4d560b1122e967bf61b61fceac43ad9ddd080"
+  },
+  "email": "chris920325@gmail.com",
+  "root": "/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2",
+  "host": "A6000-2",
+  "username": "chyang",
+  "executable": "/home/chyang/miniconda3/envs/llm-bc/bin/python3",
+  "codePathLocal": "train.py",
+  "cpu_count": 12,
+  "cpu_count_logical": 24,
+  "gpu": "NVIDIA RTX A6000",
+  "gpu_count": 2,
+  "disk": {
+    "/": {
+      "total": "1967317549056",
+      "used": "731233558528"
+    }
+  },
+  "memory": {
+    "total": "134538502144"
+  },
+  "cpu": {
+    "count": 12,
+    "countLogical": 24
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA RTX A6000",
+      "memoryTotal": "51527024640",
+      "cudaCores": 10752,
+      "architecture": "Ampere"
+    },
+    {
+      "name": "NVIDIA RTX A6000",
+      "memoryTotal": "51527024640",
+      "cudaCores": 10752,
+      "architecture": "Ampere"
+    }
+  ],
+  "cudaVersion": "12.6"
+}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_runtime":68413.573929076,"eval/steps_per_second":0.979,"train/learning_rate":0,"train_runtime":68363.1085,"train/global_step":5880,"eval/loss":0.426089346408844,"train/grad_norm":7.623011589050293,"eval/samples_per_second":31.043,"train/loss":1.6558,"train_steps_per_second":0.086,"_wandb":{"runtime":68413},"train_samples_per_second":11.01,"total_flos":3.7409731557159076e+18,"train_loss":1.6857357866352514,"train/epoch":9.984275393115173,"_step":6468,"_timestamp":1.773912674774071e+09,"eval/runtime":24.5144}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-core.log

@@ -0,0 +1,16 @@
+{"time":"2026-03-18T22:31:00.544445136+08:00","level":"INFO","msg":"started logging, with flags","port-filename":"/tmp/tmpo9n3ruuy/port-2041405.txt","pid":2041405,"debug":false,"disable-analytics":false}
+{"time":"2026-03-18T22:31:00.544808702+08:00","level":"INFO","msg":"FeatureState","shutdownOnParentExitEnabled":false}
+{"time":"2026-03-18T22:31:00.546269629+08:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":2041405}
+{"time":"2026-03-18T22:31:00.546624189+08:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":45249,"Zone":""}}
+{"time":"2026-03-18T22:31:00.712094443+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:54638"}
+{"time":"2026-03-18T22:31:01.204499962+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"x1m9280b","id":"127.0.0.1:54638"}
+{"time":"2026-03-18T22:31:01.312370477+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"x1m9280b","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:24.585911347+08:00","level":"INFO","msg":"handleInformFinish: finish message received","streamId":"x1m9280b","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:24.586245389+08:00","level":"INFO","msg":"handleInformFinish: stream closed","streamId":"x1m9280b","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077239416+08:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077301528+08:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077315957+08:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-03-19T17:31:25.077382781+08:00","level":"INFO","msg":"connection: Close: initiating connection closure","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077537714+08:00","level":"INFO","msg":"connection: Close: connection successfully closed","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077594017+08:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"127.0.0.1:54638"}
+{"time":"2026-03-19T17:31:25.077618029+08:00","level":"INFO","msg":"server is closed"}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-internal.log

@@ -0,0 +1,19 @@
+{"time":"2026-03-18T22:31:01.204991579+08:00","level":"INFO","msg":"using version","core version":"0.18.6"}
+{"time":"2026-03-18T22:31:01.205010756+08:00","level":"INFO","msg":"created symlink","path":"/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-core.log"}
+{"time":"2026-03-18T22:31:01.312346847+08:00","level":"INFO","msg":"created new stream","id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312367301+08:00","level":"INFO","msg":"stream: started","id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312665109+08:00","level":"INFO","msg":"sender: started","stream_id":"x1m9280b"}
+{"time":"2026-03-18T22:31:01.312430257+08:00","level":"INFO","msg":"handler: started","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-18T22:31:01.312656373+08:00","level":"INFO","msg":"writer: Do: started","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-18T22:31:02.758068808+08:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2026-03-19T04:22:27.963176902+08:00","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b/file_stream"}
+{"time":"2026-03-19T05:59:47.962016684+08:00","level":"INFO","msg":"api: retrying HTTP error","status":502,"url":"https://api.wandb.ai/files/chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b/file_stream"}
+{"time":"2026-03-19T17:31:14.776552794+08:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2026-03-19T17:31:14.777008363+08:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2026-03-19T17:31:15.776985704+08:00","level":"INFO","msg":"handler: operation stats","stats":{"operations":[{"desc":"uploading wandb-summary.json","runtime_seconds":0.192680949,"progress":"515B/515B"},{"desc":"saving job artifact","runtime_seconds":0.014816469}],"total_operations":2}}
+{"time":"2026-03-19T17:31:21.307731838+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-03-19T17:31:24.586012627+08:00","level":"INFO","msg":"stream: closing","id":"x1m9280b"}
+{"time":"2026-03-19T17:31:24.586046207+08:00","level":"INFO","msg":"handler: closed","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-19T17:31:24.586077373+08:00","level":"INFO","msg":"writer: Close: closed","stream_id":{"value":"x1m9280b"}}
+{"time":"2026-03-19T17:31:24.586176871+08:00","level":"INFO","msg":"sender: closed","stream_id":"x1m9280b"}
+{"time":"2026-03-19T17:31:24.586233462+08:00","level":"INFO","msg":"stream: closed","id":"x1m9280b"}

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug.log

@@ -0,0 +1,35 @@
+2026-03-18 22:31:01,198 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Current SDK version is 0.18.6
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Configure stats pid to 2041405
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from /home/chyang/.config/wandb/settings
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from /tmp2/chyang/workspace/LLM-BC/wandb/settings
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Loading settings from environment variables: {}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Applying setup settings: {'mode': 'online', '_disable_service': None}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Inferring run settings from compute environment: {'program_relpath': 'train.py', 'program_abspath': '/tmp2/chyang/workspace/LLM-BC/train.py', 'program': '/tmp2/chyang/workspace/LLM-BC/./train.py'}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_setup.py:_flush():79] Applying login settings: {}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:_log_setup():533] Logging user logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug.log
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:_log_setup():534] Logging internal logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/logs/debug-internal.log
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():619] calling init triggers
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():626] wandb.init called with sweep_config: {}
+config: {'name': 'train_llm_lowdim', '_target_': 'llmbc.workspace.train_llm_workspace.TrainLLMWorkspace', 'obs_dim': 9, 'action_dim': 4, 'horizon': 1, 'n_obs_steps': 1, 'n_action_steps': 1, 'task_name': 'box-close-v2', 'exp_name': 'train llm', 'model_name': 'mistralai/Mistral-7B-Instruct-v0.3', 'use_quantization': True, 'lora_config': {'r': 8, 'lora_alpha': 16, 'lora_dropout': 0.05, 'bias': 'none', 'task_type': 'CAUSAL_LM'}, 'dataset': {'test_data_ratio': 0.01}, 'debug': False, 'training': {'seed': 42, 'per_device_train_batch_size': 32, 'per_device_eval_batch_size': 32, 'gradient_accumulation_steps': 4, 'optim': 'paged_adamw_32bit', 'num_train_epochs': 10, 'eval_strategy': 'steps', 'logging_steps': 1, 'warmup_steps': 1000, 'logging_strategy': 'steps', 'learning_rate': 1e-05, 'fp16': False, 'bf16': True, 'tf32': True, 'group_by_length': True, 'report_to': 'wandb', 'save_steps': 5000, 'eval_steps': 10, 'use_joint_mlp_projector': True, 'joint_obs_action_mlp_lr': 5e-06}, 'trainer': {'obs_dim': 9, 'action_dim': 4, 'use_joint_mlp_projector': True, 'max_seq_length': 100, 'dataset_text_field': 'text', 'packing': False}, 'logging': {'project': 'llm_module_finetuning', 'resume': True, 'mode': 'online', 'name': '2026.03.18-22.30.56_train_llm_lowdim_box-close-v2', 'tags': ['train_llm_lowdim', 'box-close-v2', 'train llm'], 'id': None, 'group': None}, 'multi_run': {'run_dir': 'data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2', 'wandb_name_base': '2026.03.18-22.30.56_train_llm_lowdim_box-close-v2'}, 'task': {'name': 'box-close-v2', 'obs_dim': 9, 'action_dim': 4, 'env_runner': {'_target_': 'llmbc.env_runner.metaworld_lowdim_runner.MetaworldLowdimRunner', 'env_name': 'llf-metaworld-box-close-v2', 'n_train': 10, 'n_test': 50, 'n_envs': 10, 'max_steps': 30, 'n_obs_steps': 1, 'n_action_steps': 1, 'instruction_type': 'b', 'feedback_type': ['hp', 'hn', 'fp'], 'visual': False, 'discount': 0.9}, 'dataset': {'_target_': 'llmbc.dataset.metaworld_lowdim_dataset.MetaworldLowdimDataset', 'data_path': 'datasets/box-close-v2-general.pt', 'data_path2': 'datasets/box-close-v2.pt', 'horizon': 1, 'pad_before': 0, 'pad_after': 0, 'obs_eef_target': True, 'use_manual_normalizer': False, 'val_ratio': 0.05, 'dummy_normalizer': False}, 'instructor': {'_target_': 'llmbc.translator.instructor.metaworld_instructor.box_close_v2_instructor.BoxCloseV2Instructor'}}, 'llm': {'name': 'mistralai/Mistral-7B-Instruct-v0.3', 'model_name': 'Mistral-7B-Instruct-v0.3', 'config_target': 'llmbc.model.llm.mistral_lowdim_model.LowdimMistralConfig', 'causal_lm_target': 'llmbc.model.llm.mistral_lowdim_model.LowdimMistralForCausalLM', 'use_quantization': True, 'use_joint_mlp_projector': True, 'llm_mode': 'mlp-finetuned', 'finetune_mode': 'lora', 'checkpoint': 'data/outputs/2026.03.14/23.40.55_train_mlp_projector_box-close-v2/checkpoints/latest.ckpt', 'max_length': 100, 'lora_config': {'r': 8, 'lora_alpha': 16, 'lora_dropout': 0.05, 'bias': 'none', 'task_type': 'CAUSAL_LM'}, 'prompter': {'_target_': 'llmbc.translator.prompter.mistral_prompter.MistralPrompter', 'use_joint_mlp_projector': True}, 'hydra': {'job': {'override_dirname': 'mistralai/Mistral-7B-Instruct-v0.3'}, 'run': {'dir': 'data/outputs/2026.03.18/22.30.56_mistralai/Mistral-7B-Instruct-v0.3'}}}}
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():669] starting backend
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [wandb_init.py:init():673] sending inform_init request
+2026-03-18 22:31:01,199 INFO    MainThread:2041405 [backend.py:_multiprocessing_setup():104] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2026-03-18 22:31:01,200 INFO    MainThread:2041405 [wandb_init.py:init():686] backend started and connected
+2026-03-18 22:31:01,204 INFO    MainThread:2041405 [wandb_init.py:init():781] updated telemetry
+2026-03-18 22:31:01,224 INFO    MainThread:2041405 [wandb_init.py:init():814] communicating run to backend with 90.0 second timeout
+2026-03-18 22:31:02,754 INFO    MainThread:2041405 [wandb_init.py:init():867] starting run threads in backend
+2026-03-18 22:31:02,923 INFO    MainThread:2041405 [wandb_run.py:_console_start():2451] atexit reg
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2299] redirect: wrap_raw
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2364] Wrapping output streams.
+2026-03-18 22:31:02,924 INFO    MainThread:2041405 [wandb_run.py:_redirect():2389] Redirects installed.
+2026-03-18 22:31:02,925 INFO    MainThread:2041405 [wandb_init.py:init():911] run started, returning control to user process
+2026-03-18 22:31:51,682 INFO    MainThread:2041405 [wandb_run.py:_config_callback():1389] config_cb None None {'peft_config': {'default': {'task_type': 'CAUSAL_LM', 'peft_type': <PeftType.LORA: 'LORA'>, 'auto_mapping': None, 'base_model_name_or_path': 'mistralai/Mistral-7B-Instruct-v0.3', 'revision': None, 'inference_mode': False, 'r': 8, 'target_modules': {'down_proj', 'v_proj', 'k_proj', 'lm_head', 'up_proj', 'q_proj', 'gate_proj', 'o_proj'}, 'exclude_modules': None, 'lora_alpha': 16, 'lora_dropout': 0.05, 'fan_in_fan_out': False, 'bias': 'none', 'use_rslora': False, 'modules_to_save': None, 'init_lora_weights': True, 'layers_to_transform': None, 'layers_pattern': None, 'rank_pattern': {}, 'alpha_pattern': {}, 'megatron_config': None, 'megatron_core': 'megatron.core', 'loftq_config': {}, 'eva_config': None, 'use_dora': False, 'layer_replication': None, 'runtime_config': {'ephemeral_gpu_offload': False}, 'lora_bias': False}}, 'obs_dim': 9, 'action_dim': 4, 'use_joint_mlp_projector': True, 'vocab_size': 32768, 'max_position_embeddings': 32768, 'hidden_size': 4096, 'intermediate_size': 14336, 'num_hidden_layers': 32, 'num_attention_heads': 32, 'sliding_window': None, 'head_dim': 128, 'num_key_value_heads': 8, 'hidden_act': 'silu', 'initializer_range': 0.02, 'rms_norm_eps': 1e-05, 'use_cache': False, 'rope_theta': 1000000.0, 'attention_dropout': 0.0, 'return_dict': True, 'output_hidden_states': False, 'output_attentions': False, 'torchscript': False, 'torch_dtype': 'bfloat16', 'use_bfloat16': False, 'tf_legacy_loss': False, 'pruned_heads': {}, 'tie_word_embeddings': False, 'chunk_size_feed_forward': 0, 'is_encoder_decoder': False, 'is_decoder': False, 'cross_attention_hidden_size': None, 'add_cross_attention': False, 'tie_encoder_decoder': False, 'max_length': 20, 'min_length': 0, 'do_sample': False, 'early_stopping': False, 'num_beams': 1, 'num_beam_groups': 1, 'diversity_penalty': 0.0, 'temperature': 1.0, 'top_k': 50, 'top_p': 1.0, 'typical_p': 1.0, 'repetition_penalty': 1.0, 'length_penalty': 1.0, 'no_repeat_ngram_size': 0, 'encoder_no_repeat_ngram_size': 0, 'bad_words_ids': None, 'num_return_sequences': 1, 'output_scores': False, 'return_dict_in_generate': False, 'forced_bos_token_id': None, 'forced_eos_token_id': None, 'remove_invalid_values': False, 'exponential_decay_length_penalty': None, 'suppress_tokens': None, 'begin_suppress_tokens': None, 'architectures': ['MistralForCausalLM'], 'finetuning_task': None, 'id2label': {0: 'LABEL_0', 1: 'LABEL_1'}, 'label2id': {'LABEL_0': 0, 'LABEL_1': 1}, 'tokenizer_class': None, 'prefix': None, 'bos_token_id': 1, 'pad_token_id': None, 'eos_token_id': 2, 'sep_token_id': None, 'decoder_start_token_id': None, 'task_specific_params': None, 'problem_type': None, '_name_or_path': 'mistralai/Mistral-7B-Instruct-v0.3', '_attn_implementation_autoset': True, 'transformers_version': '4.47.1', 'model_type': 'mistral_lowdim', 'quantization_config': {'quant_method': 'BITS_AND_BYTES', '_load_in_8bit': False, '_load_in_4bit': True, 'llm_int8_threshold': 6.0, 'llm_int8_skip_modules': ['joint_obs_action_projector'], 'llm_int8_enable_fp32_cpu_offload': False, 'llm_int8_has_fp16_weight': False, 'bnb_4bit_quant_type': 'nf4', 'bnb_4bit_use_double_quant': True, 'bnb_4bit_compute_dtype': 'bfloat16', 'bnb_4bit_quant_storage': 'uint8', 'load_in_4bit': True, 'load_in_8bit': False}, 'output_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2', 'overwrite_output_dir': False, 'do_train': False, 'do_eval': True, 'do_predict': False, 'eval_strategy': 'steps', 'prediction_loss_only': False, 'per_device_train_batch_size': 32, 'per_device_eval_batch_size': 32, 'per_gpu_train_batch_size': None, 'per_gpu_eval_batch_size': None, 'gradient_accumulation_steps': 4, 'eval_accumulation_steps': None, 'eval_delay': 0, 'torch_empty_cache_steps': None, 'learning_rate': 1e-05, 'weight_decay': 0.0, 'adam_beta1': 0.9, 'adam_beta2': 0.999, 'adam_epsilon': 1e-08, 'max_grad_norm': 1.0, 'num_train_epochs': 10, 'max_steps': -1, 'lr_scheduler_type': 'linear', 'lr_scheduler_kwargs': {}, 'warmup_ratio': 0.0, 'warmup_steps': 1000, 'log_level': 'passive', 'log_level_replica': 'warning', 'log_on_each_node': True, 'logging_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2/runs/Mar18_22-31-41_A6000-2', 'logging_strategy': 'steps', 'logging_first_step': False, 'logging_steps': 1, 'logging_nan_inf_filter': True, 'save_strategy': 'steps', 'save_steps': 5000, 'save_total_limit': None, 'save_safetensors': True, 'save_on_each_node': False, 'save_only_model': False, 'restore_callback_states_from_checkpoint': False, 'no_cuda': False, 'use_cpu': False, 'use_mps_device': False, 'seed': 42, 'data_seed': None, 'jit_mode_eval': False, 'use_ipex': False, 'bf16': True, 'fp16': False, 'fp16_opt_level': 'O1', 'half_precision_backend': 'auto', 'bf16_full_eval': False, 'fp16_full_eval': False, 'tf32': True, 'local_rank': 0, 'ddp_backend': None, 'tpu_num_cores': None, 'tpu_metrics_debug': False, 'debug': [], 'dataloader_drop_last': False, 'eval_steps': 10, 'dataloader_num_workers': 0, 'dataloader_prefetch_factor': None, 'past_index': -1, 'run_name': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/mistralai/Mistral-7B-Instruct-v0.3-finetuned-box-close-v2', 'disable_tqdm': False, 'remove_unused_columns': True, 'label_names': None, 'load_best_model_at_end': False, 'metric_for_best_model': None, 'greater_is_better': None, 'ignore_data_skip': False, 'fsdp': [], 'fsdp_min_num_params': 0, 'fsdp_config': {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, 'fsdp_transformer_layer_cls_to_wrap': None, 'accelerator_config': {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}, 'deepspeed': None, 'label_smoothing_factor': 0.0, 'optim': 'paged_adamw_32bit', 'optim_args': None, 'adafactor': False, 'group_by_length': True, 'length_column_name': 'length', 'report_to': ['wandb'], 'ddp_find_unused_parameters': None, 'ddp_bucket_cap_mb': None, 'ddp_broadcast_buffers': None, 'dataloader_pin_memory': True, 'dataloader_persistent_workers': False, 'skip_memory_metrics': True, 'use_legacy_prediction_loop': False, 'push_to_hub': False, 'resume_from_checkpoint': None, 'hub_model_id': None, 'hub_strategy': 'every_save', 'hub_token': '<HUB_TOKEN>', 'hub_private_repo': None, 'hub_always_push': False, 'gradient_checkpointing': False, 'gradient_checkpointing_kwargs': None, 'include_inputs_for_metrics': False, 'include_for_metrics': [], 'eval_do_concat_batches': True, 'fp16_backend': 'auto', 'evaluation_strategy': None, 'push_to_hub_model_id': None, 'push_to_hub_organization': None, 'push_to_hub_token': '<PUSH_TO_HUB_TOKEN>', 'mp_parameters': '', 'auto_find_batch_size': False, 'full_determinism': False, 'torchdynamo': None, 'ray_scope': 'last', 'ddp_timeout': 1800, 'torch_compile': False, 'torch_compile_backend': None, 'torch_compile_mode': None, 'dispatch_batches': None, 'split_batches': None, 'include_tokens_per_second': False, 'include_num_input_tokens_seen': False, 'neftune_noise_alpha': None, 'optim_target_modules': None, 'batch_eval_metrics': False, 'eval_on_start': False, 'use_liger_kernel': False, 'eval_use_gather_object': False, 'average_tokens_across_devices': False, 'dataset_text_field': 'text', 'packing': False, 'max_seq_length': 100, 'dataset_num_proc': None, 'dataset_batch_size': 1000, 'model_init_kwargs': None, 'dataset_kwargs': {}, 'eval_packing': None, 'num_of_sequences': 1024, 'chars_per_token': '<CHARS_PER_TOKEN>', 'use_liger': False, 'joint_obs_action_mlp_lr': 5e-06, 'obs_mlp_lr': None, 'action_mlp_lr': None}
+2026-03-18 22:31:51,686 INFO    MainThread:2041405 [wandb_config.py:__setitem__():154] config set model/num_parameters = 7286128640 - <bound method Run._config_callback of <wandb.sdk.wandb_run.Run object at 0x7fe2561236a0>>
+2026-03-18 22:31:51,686 INFO    MainThread:2041405 [wandb_run.py:_config_callback():1389] config_cb model/num_parameters 7286128640 None
+2026-03-19 17:31:14,775 INFO    MainThread:2041405 [wandb_run.py:_finish():2146] finishing run chyang25-national-taiwan-university/llm_module_finetuning/x1m9280b
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_atexit_cleanup():2414] got exitcode: 0
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_restore():2396] restore
+2026-03-19 17:31:14,776 INFO    MainThread:2041405 [wandb_run.py:_restore():2402] restore done
+2026-03-19 17:31:24,580 INFO    MainThread:2041405 [wandb_run.py:_footer_history_summary_info():3963] rendering history
+2026-03-19 17:31:24,581 INFO    MainThread:2041405 [wandb_run.py:_footer_history_summary_info():3995] rendering summary
+2026-03-19 17:31:24,585 INFO    MainThread:2041405 [wandb_run.py:_footer_sync_info():3922] logging synced files

2026.03.18/22.30.56_train_llm_lowdim_box-close-v2/wandb/run-20260318_223101-x1m9280b/run-x1m9280b.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28702b88803b5c2c0e3359c7dded6327484703469450657739c49f1cb5b68fbe
+size 35538433