Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+GSM8k/Qwen2.5-1.5B-Instruct_balanced_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+GSM8k/Qwen2.5-1.5B-Instruct_cosine_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+GSM8k/Qwen2.5-1.5B-Instruct_gaussian_dapo/tokenizer.json filter=lfs diff=lfs merge=lfs -text

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/.hydra/config.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: balanced
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/.hydra/hydra.yaml

@@ -0,0 +1,171 @@
+hydra:
+  run:
+    dir: ${output.root_path}/outputs/${mode2name:${mode},${output.run_name},${model.trim}}
+  sweep:
+    dir: ${output.root_path}/multirun/${now:%Y%m%d}
+    subdir: ${hydra.job.override_dirname}
+  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:
+    - mode=train
+    - model=qwen15
+    - task=aqua
+    - algorithm=grpo
+    - algorithm.training.curriculum_schedule=balanced
+    - algorithm.training.scheduler_params.mu_exp=0.5
+    - algorithm.training.scheduler_params.sigma=0.5
+    - algorithm.training.max_steps=1600
+    - algorithm.training.curriculum=false
+    - algorithm.training.vllm_gpu_memory_utilization=0.8
+    - algorithm.training.vllm_mode=server
+    - algorithm.training.report_to=[tensorboard]
+    - algorithm.training.push_to_hub=false
+    - algorithm.training.save_strategy=no
+  job:
+    name: main
+    chdir: false
+    override_dirname: algorithm.training.curriculum=false,algorithm.training.curriculum_schedule=balanced,algorithm.training.max_steps=1600,algorithm.training.push_to_hub=false,algorithm.training.report_to=[tensorboard],algorithm.training.save_strategy=no,algorithm.training.scheduler_params.mu_exp=0.5,algorithm.training.scheduler_params.sigma=0.5,algorithm.training.vllm_gpu_memory_utilization=0.8,algorithm.training.vllm_mode=server,algorithm=grpo,mode=train,model=qwen15,task=aqua
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /scratch/10416/sushil22/projects/Sys2Bench_shurui
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /scratch/10416/sushil22/projects/Sys2Bench_shurui/methods/RL/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /scratch/10416/sushil22/projects/Sys2Bench_shurui/outputs/Qwen2.5-1.5B-Instruct_aqua_grpo_balanced_0.5_0.5_SEC0.3DRO1.0G0.0_minpTrue_1600
+    choices:
+      algorithm: grpo
+      task: aqua
+      model: qwen15
+      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

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/.hydra/overrides.yaml

@@ -0,0 +1,14 @@
+- mode=train
+- model=qwen15
+- task=aqua
+- algorithm=grpo
+- algorithm.training.curriculum_schedule=balanced
+- algorithm.training.scheduler_params.mu_exp=0.5
+- algorithm.training.scheduler_params.sigma=0.5
+- algorithm.training.max_steps=1600
+- algorithm.training.curriculum=false
+- algorithm.training.vllm_gpu_memory_utilization=0.8
+- algorithm.training.vllm_mode=server
+- algorithm.training.report_to=[tensorboard]
+- algorithm.training.push_to_hub=false
+- algorithm.training.save_strategy=no

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/chat_template.jinja

@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/config.json

@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/config_dump.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: balanced
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/events.out.tfevents.1763473963.c317-013.ls6.tacc.utexas.edu.3083110.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce72c3fc0e685e256e990eba01a7a628cbc0c122c8aa7b20b02c221bfc6745d7
+size 240333

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.53.1"
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5eec44635b7f15a25f77cce44319760a6ca0a96bcf2149cf415c820299e7e2f
+size 3087467144

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/test_results.json

@@ -0,0 +1,27 @@
+{
+    "overall": {
+        "avg_reward": 0.6889763779527559,
+        "accuracy": 0.65748031496063,
+        "support": 254
+    },
+    "trivial": {
+        "avg_reward": 0.9500000000000001,
+        "accuracy": 0.9444444444444444,
+        "support": 72
+    },
+    "easy": {
+        "avg_reward": 0.75,
+        "accuracy": 0.7222222222222222,
+        "support": 72
+    },
+    "medium": {
+        "avg_reward": 0.5333333333333333,
+        "accuracy": 0.4861111111111111,
+        "support": 72
+    },
+    "hard": {
+        "avg_reward": 0.37368421052631573,
+        "accuracy": 0.3157894736842105,
+        "support": 38
+    }
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eee858c5123a4279c3e1f7b81247343f356ac767940b2692a928ad929543214
+size 11422063

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

AQUA/Qwen2.5-1.5B-Instruct_balanced_dapo/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f7ebef5aa32f22a99a127e32b5781009bdbedaee8d39ba73e484773759c49ce
+size 8056

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/.hydra/config.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: cosine
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/.hydra/hydra.yaml

@@ -0,0 +1,171 @@
+hydra:
+  run:
+    dir: ${output.root_path}/outputs/${mode2name:${mode},${output.run_name},${model.trim}}
+  sweep:
+    dir: ${output.root_path}/multirun/${now:%Y%m%d}
+    subdir: ${hydra.job.override_dirname}
+  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:
+    - mode=train
+    - model=qwen15
+    - task=aqua
+    - algorithm=grpo
+    - algorithm.training.curriculum_schedule=cosine
+    - algorithm.training.scheduler_params.mu_exp=0.5
+    - algorithm.training.scheduler_params.sigma=0.5
+    - algorithm.training.max_steps=1600
+    - algorithm.training.curriculum=false
+    - algorithm.training.vllm_gpu_memory_utilization=0.8
+    - algorithm.training.vllm_mode=server
+    - algorithm.training.report_to=[tensorboard]
+    - algorithm.training.push_to_hub=false
+    - algorithm.training.save_strategy=no
+  job:
+    name: main
+    chdir: false
+    override_dirname: algorithm.training.curriculum=false,algorithm.training.curriculum_schedule=cosine,algorithm.training.max_steps=1600,algorithm.training.push_to_hub=false,algorithm.training.report_to=[tensorboard],algorithm.training.save_strategy=no,algorithm.training.scheduler_params.mu_exp=0.5,algorithm.training.scheduler_params.sigma=0.5,algorithm.training.vllm_gpu_memory_utilization=0.8,algorithm.training.vllm_mode=server,algorithm=grpo,mode=train,model=qwen15,task=aqua
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /scratch/10416/sushil22/projects/Sys2Bench_shurui
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /scratch/10416/sushil22/projects/Sys2Bench_shurui/methods/RL/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /scratch/10416/sushil22/projects/Sys2Bench_shurui/outputs/Qwen2.5-1.5B-Instruct_aqua_grpo_cosine_0.5_0.5_SEC0.3DRO1.0G0.0_minpTrue_1600
+    choices:
+      algorithm: grpo
+      task: aqua
+      model: qwen15
+      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

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/.hydra/overrides.yaml

@@ -0,0 +1,14 @@
+- mode=train
+- model=qwen15
+- task=aqua
+- algorithm=grpo
+- algorithm.training.curriculum_schedule=cosine
+- algorithm.training.scheduler_params.mu_exp=0.5
+- algorithm.training.scheduler_params.sigma=0.5
+- algorithm.training.max_steps=1600
+- algorithm.training.curriculum=false
+- algorithm.training.vllm_gpu_memory_utilization=0.8
+- algorithm.training.vllm_mode=server
+- algorithm.training.report_to=[tensorboard]
+- algorithm.training.push_to_hub=false
+- algorithm.training.save_strategy=no

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/chat_template.jinja

@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/config.json

@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/config_dump.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: cosine
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/events.out.tfevents.1763473040.c317-015.ls6.tacc.utexas.edu.819015.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ababb120bf709b17124d5b0dad3e1f4b160699f3deab95fadca314a128bdaca9
+size 240325

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.53.1"
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec25a67197d1ba53ea579e5c0305dc975425bc8444e422693f560af433855f52
+size 3087467144

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/test_results.json

@@ -0,0 +1,27 @@
+{
+    "overall": {
+        "avg_reward": 0.7240157480314959,
+        "accuracy": 0.6968503937007874,
+        "support": 254
+    },
+    "trivial": {
+        "avg_reward": 0.9625000000000001,
+        "accuracy": 0.9583333333333334,
+        "support": 72
+    },
+    "easy": {
+        "avg_reward": 0.7722222222222221,
+        "accuracy": 0.75,
+        "support": 72
+    },
+    "medium": {
+        "avg_reward": 0.5944444444444446,
+        "accuracy": 0.5555555555555556,
+        "support": 72
+    },
+    "hard": {
+        "avg_reward": 0.42631578947368426,
+        "accuracy": 0.3684210526315789,
+        "support": 38
+    }
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eee858c5123a4279c3e1f7b81247343f356ac767940b2692a928ad929543214
+size 11422063

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/tokenizer_config.json

@@ -0,0 +1,207 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

AQUA/Qwen2.5-1.5B-Instruct_cosine_dapo/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82d31421b25b5a50b5ba38fb66528f5df3e5fe731f5b6720b48f37d57418df96
+size 8056

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/.hydra/config.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: gaussian
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/.hydra/hydra.yaml

@@ -0,0 +1,171 @@
+hydra:
+  run:
+    dir: ${output.root_path}/outputs/${mode2name:${mode},${output.run_name},${model.trim}}
+  sweep:
+    dir: ${output.root_path}/multirun/${now:%Y%m%d}
+    subdir: ${hydra.job.override_dirname}
+  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:
+    - mode=train
+    - model=qwen15
+    - task=aqua
+    - algorithm=grpo
+    - algorithm.training.curriculum_schedule=gaussian
+    - algorithm.training.scheduler_params.mu_exp=0.5
+    - algorithm.training.scheduler_params.sigma=0.5
+    - algorithm.training.max_steps=1600
+    - algorithm.training.curriculum=false
+    - algorithm.training.vllm_gpu_memory_utilization=0.8
+    - algorithm.training.vllm_mode=server
+    - algorithm.training.report_to=[tensorboard]
+    - algorithm.training.push_to_hub=false
+    - algorithm.training.save_strategy=no
+  job:
+    name: main
+    chdir: false
+    override_dirname: algorithm.training.curriculum=false,algorithm.training.curriculum_schedule=gaussian,algorithm.training.max_steps=1600,algorithm.training.push_to_hub=false,algorithm.training.report_to=[tensorboard],algorithm.training.save_strategy=no,algorithm.training.scheduler_params.mu_exp=0.5,algorithm.training.scheduler_params.sigma=0.5,algorithm.training.vllm_gpu_memory_utilization=0.8,algorithm.training.vllm_mode=server,algorithm=grpo,mode=train,model=qwen15,task=aqua
+    id: ???
+    num: ???
+    config_name: config
+    env_set: {}
+    env_copy: []
+    config:
+      override_dirname:
+        kv_sep: '='
+        item_sep: ','
+        exclude_keys: []
+  runtime:
+    version: 1.3.2
+    version_base: '1.3'
+    cwd: /scratch/10416/sushil22/projects/Sys2Bench_shurui
+    config_sources:
+    - path: hydra.conf
+      schema: pkg
+      provider: hydra
+    - path: /scratch/10416/sushil22/projects/Sys2Bench_shurui/methods/RL/conf
+      schema: file
+      provider: main
+    - path: ''
+      schema: structured
+      provider: schema
+    output_dir: /scratch/10416/sushil22/projects/Sys2Bench_shurui/outputs/Qwen2.5-1.5B-Instruct_aqua_grpo_gaussian_0.5_0.5_SEC0.3DRO1.0G0.0_minpTrue_1600
+    choices:
+      algorithm: grpo
+      task: aqua
+      model: qwen15
+      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

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/.hydra/overrides.yaml

@@ -0,0 +1,14 @@
+- mode=train
+- model=qwen15
+- task=aqua
+- algorithm=grpo
+- algorithm.training.curriculum_schedule=gaussian
+- algorithm.training.scheduler_params.mu_exp=0.5
+- algorithm.training.scheduler_params.sigma=0.5
+- algorithm.training.max_steps=1600
+- algorithm.training.curriculum=false
+- algorithm.training.vllm_gpu_memory_utilization=0.8
+- algorithm.training.vllm_mode=server
+- algorithm.training.report_to=[tensorboard]
+- algorithm.training.push_to_hub=false
+- algorithm.training.save_strategy=no

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/added_tokens.json

@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/chat_template.jinja

@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/config.json

@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/config_dump.yaml

@@ -0,0 +1,85 @@
+mode: train
+experiment:
+  dataset_size: 6000
+  dataset_seed: 1234
+  test_size: 0.1
+  hf_token: ${oc.env:HF_TOKEN,null}
+output:
+  root_path: ${oc.env:ROOT_PATH}
+  run_name: ${model.trim}_${task.name}_${algorithm.name}_${algorithm.training.curriculum_schedule}_${algorithm.training.scheduler_params.mu_exp}_${algorithm.training.scheduler_params.sigma}_SEC${algorithm.training.scheduler_params.vrex_adds.sec}DRO${algorithm.training.scheduler_params.vrex_adds.groupdro}G${algorithm.training.scheduler_params.vrex_adds.gaussian}_minp${algorithm.training.scheduler_params.min_prob}${ckpt2short:${algorithm.training.resume_from_checkpoint}}_${algorithm.training.max_steps}
+lora:
+  r: 32
+  alpha: 64
+  dropout: 0.1
+  target_modules:
+  - q_proj
+  - v_proj
+  task_type: CAUSAL_LM
+occupy_gpu_memory: false
+occupy_gpu_memory_gb: 50
+gpu_device: cuda:0
+model:
+  family: Qwen
+  trim: Qwen2.5-1.5B-Instruct
+  name: ${model.family}/${model.trim}
+  trust_remote_code: true
+  torch_dtype: bfloat16
+  attn_implementation: flash_attention_2
+task:
+  name: aqua
+  training:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    max_prompt_length: 512
+    max_completion_length: 512
+  inference:
+    data_files:
+    - datasets/aqua/trivial
+    - datasets/aqua/easy
+    - datasets/aqua/medium
+    - datasets/aqua/hard
+    temperature: 0
+    'n': 1
+algorithm:
+  name: grpo
+  training:
+    resume_from_checkpoint: null
+    learning_rate: 1.0e-06
+    lr_scheduler_type: cosine
+    logging_steps: 10
+    max_steps: 1600
+    per_device_train_batch_size: 16
+    generation_batch_size: null
+    steps_per_generation: 1
+    gradient_accumulation_steps: 4
+    gradient_checkpointing: true
+    bf16: true
+    report_to:
+    - tensorboard
+    push_to_hub: false
+    save_strategy: 'no'
+    save_steps: ${algorithm.training.max_steps}
+    tf32: true
+    num_generations: 8
+    beta: 0.001
+    use_vllm: true
+    vllm_mode: server
+    vllm_gpu_memory_utilization: 0.8
+    vllm_server_port: 8000
+    curriculum: false
+    curriculum_schedule: gaussian
+    scheduler_params:
+      mu_exp: 0.5
+      sigma: 0.5
+      vrex_adds:
+        groupdro: 1.0
+        gaussian: 0.0
+        sec: 0.3
+      beta: 1.0
+      min_prob: true
+      td_alpha: 0.5
+      sec_temperature: 0.3
+      max_dapo_iter: 4

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/events.out.tfevents.1763472877.c316-011.ls6.tacc.utexas.edu.1952186.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5fbbc93a5cafde598633aba3331d14dd72555de9d0805cecdf9c1f658af1aeb
+size 240333

AQUA/Qwen2.5-1.5B-Instruct_gaussian_dapo/generation_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_k": 20,
+  "top_p": 0.8,
+  "transformers_version": "4.53.1"
+}