Tinker 后端

Tinker 后端#

备注

本示例演示了如何在 Trinity-RFT 中使用 Tinker，从而在无 GPU的设备上进行模型训练。

安装与配置#

1. API Key 配置#

在启动 Ray 之前，必须将 TRINITY_API_KEY 环境变量设置为你的 Tinker API 密钥，以便正确访问 Tinker 的 API：

export TRINITY_API_KEY=your_tinker_api_key
ray start --head

2. 配置文件#

在 YAML 配置文件中通过如下方式设置 model.tinker 参数以启用 Tinker 后端：

model:
  tinker:
    enable: true
    base_model: null
    rank: 32
    seed: null
    train_mlp: true
    train_attn: true
    train_unembed: true

配置参数说明#

  • tinker：Tinker 专用配置部分。注意：启用 Tinker 后，所有 LoRA 配置（model.lora_configs）将被忽略。

    • enable：是否启用 Tinker 后端。默认值：false

    • base_model：Tinker 的基础模型路径。如果未指定（null），则默认为配置中其他位置的 model_path

    • rank：LoRA 的秩，控制适应矩阵的大小。默认值：32

    • seed：Tinker 操作的随机种子。未指定（null）时不设定特定种子

    • train_mlp：是否训练 MLP（前馈）层。默认值：true

    • train_attn：是否训练注意力层。默认值：true

    • train_unembed：是否训练输出（unembedding）层。默认值：true

使用方法#

配置完成后，Trinity-RFT 使用 Tinker 后端的方式与标准 veRL 后端一致。启动训练命令如下：

trinity run --config tinker.yaml  # 请替换为你的实际配置文件路径

Tinker 后端的功能限制#

  1. 熵损失（entropy loss） 与 veRL 后端不完全一致。

  2. 不支持 compute_advantage_in_trainer=true 的算法，包括：

    • PPO（algorithm.algorithm_type=ppo）

    • Reinforce++（algorithm.algorithm_type=reinforceplusplus）

    • RLOO（algorithm.algorithm_type=rloo）

    • On-policy distillation（algorithm.algorithm_type=on_policy_distill）

    目前支持 grpo, opmd, sft 等算法，未来会支持更多算法。

  3. 暂不支持多阶段训练，后续会添加该功能。

💡 完整的示例配置文件见 tinker.yaml。

Llama-3.2-3B 模型实验结果#

我们在 GSM8K 数据集上，分别使用 Tinker 和 veRL 后端对 Llama-3.2-3B 模型进行了训练。以下为实验中使用的完整配置文件。

点击展开：Tinker 后端配置 
mode: both
project: Trinity-RFT-gsm8k
group: alignment-tinker
name: tinker-llama3.2-3B-off1
checkpoint_root_dir: ${oc.env:TRINITY_CHECKPOINT_ROOT_DIR,./checkpoints}
algorithm:
  algorithm_type: grpo
  repeat_times: 8
  kl_loss_fn_args:
    kl_coef: 0.0
  optimizer:
    lr: 1.0e-05
model:
  model_path: meta-llama/Llama-3.2-3B
  max_prompt_tokens: 1024
  max_response_tokens: 2048
  custom_chat_template: "{{- bos_token }}\n{%- if custom_tools is defined %}\n    {%- set tools = custom_tools %}\n{%- endif %}\n{%- if not tools_in_user_message is defined %}\n    {%- set tools_in_user_message = true %}\n{%- endif %}\n{%- if not date_string is defined %}\n    {%- if strftime_now is defined %}\n        {%- set date_string = strftime_now(\"%d %b %Y\") %}\n    {%- else %}\n        {%- set date_string = \"26 Jul 2024\" %}\n    {%- endif %}\n{%- endif %}\n{%- if not tools is defined %}\n    {%- set tools = none %}\n{%- endif %}\n\n{#- This block extracts the system message, so we can slot it into the right place. #}\n{%- if messages[0]['role'] == 'system' %}\n    {%- set system_message = messages[0]['content']|trim %}\n    {%- set messages = messages[1:] %}\n{%- else %}\n    {%- set system_message = \"\" %}\n{%- endif %}\n\n{#- System message #}\n{{- \"<|start_header_id|>system<|end_header_id|>\\n\\n\" }}\n{%- if tools is not none %}\n    {{- \"Environment: ipython\\n\" }}\n{%- endif %}\n{{- \"Cutting Knowledge Date: December 2023\\n\" }}\n{{- \"Today Date: \" + date_string + \"\\n\\n\" }}\n{%- if tools is not none and not tools_in_user_message %}\n    {{- \"You have access to the following functions. To call a function, please respond with JSON for a function call.\" }}\n    {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n    {{- \"Do not use variables.\\n\\n\" }}\n    {%- for t in tools %}\n        {{- t | tojson(indent=4) }}\n        {{- \"\\n\\n\" }}\n    {%- endfor %}\n{%- endif %}\n{{- system_message }}\n{{- \"<|eot_id|>\" }}\n\n{#- Custom tools are passed in a user message with some extra guidance #}\n{%- if tools_in_user_message and not tools is none %}\n    {#- Extract the first user message so we can plug it in here #}\n    {%- if messages | length != 0 %}\n        {%- set first_user_message = messages[0]['content']|trim %}\n        {%- set messages = messages[1:] %}\n    {%- else %}\n        {{- raise_exception(\"Cannot put tools in the first user message when there's no first user message!\") }}\n{%- endif %}\n    {{- '<|start_header_id|>user<|end_header_id|>\\n\\n' -}}\n    {{- \"Given the following functions, please respond with a JSON for a function call \" }}\n    {{- \"with its proper arguments that best answers the given prompt.\\n\\n\" }}\n    {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n    {{- \"Do not use variables.\\n\\n\" }}\n    {%- for t in tools %}\n        {{- t | tojson(indent=4) }}\n        {{- \"\\n\\n\" }}\n    {%- endfor %}\n    {{- first_user_message + \"<|eot_id|>\"}}\n{%- endif %}\n\n{%- for message in messages %}\n    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}\n        {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\\n\\n'+ message['content'] | trim + '<|eot_id|>' }}\n    {%- elif 'tool_calls' in message %}\n        {%- if not message.tool_calls|length == 1 %}\n            {{- raise_exception(\"This model only supports single tool-calls at once!\") }}\n        {%- endif %}\n        {%- set tool_call = message.tool_calls[0].function %}\n        {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n        {{- '{\"name\": \"' + tool_call.name + '\", ' }}\n        {{- '\"parameters\": ' }}\n        {{- tool_call.arguments | tojson }}\n        {{- \"}\" }}\n        {{- \"<|eot_id|>\" }}\n    {%- elif message.role == \"tool\" or message.role == \"ipython\" %}\n        {{- \"<|start_header_id|>ipython<|end_header_id|>\\n\\n\" }}\n        {%- if message.content is mapping or message.content is iterable %}\n            {{- message.content | tojson }}\n        {%- else %}\n            {{- message.content }}\n        {%- endif %}\n        {{- \"<|eot_id|>\" }}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' }}\n{%- endif %}\n"
  tinker:
    enable: true
    base_model: meta-llama/Llama-3.2-3B
cluster:
  node_num: 1
  gpu_per_node: 8
buffer:
  batch_size: 96
  total_epochs: 1
  explorer_input:
    taskset:
      name: taskset
      storage_type: file
      path: openai/gsm8k
      split: train
      subset_name: main
      format:
        prompt_key: question
        response_key: answer
    default_workflow_type: math_workflow
  trainer_input:
    experience_buffer:
      name: experience_buffer
      storage_type: queue
explorer:
  runner_per_model: 16
  rollout_model:
    engine_num: 4
    seed: 42
trainer:
  save_interval: 100
  grad_clip: 1.0
monitor:
  monitor_type: wandb
synchronizer:
  sync_method: checkpoint
  sync_style: fixed
  sync_interval: 1
  sync_offset: 1
  sync_timeout: 1200
点击展开：veRL 后端配置（LoRA） 
mode: both
project: Trinity-RFT-gsm8k
group: alignment-tinker
name: verl-llama3.2-3B-lora-off1
checkpoint_root_dir: ${oc.env:TRINITY_CHECKPOINT_ROOT_DIR,./checkpoints}
algorithm:
  algorithm_type: grpo
  repeat_times: 8
  kl_loss_fn_args:
    kl_coef: 0.0
  optimizer:
    lr: 1.0e-05
data_processor: {}
model:
  model_path: meta-llama/Llama-3.2-3B
  max_prompt_tokens: 1024
  max_response_tokens: 2048
  custom_chat_template: "{{- bos_token }}\n{%- if custom_tools is defined %}\n    {%- set tools = custom_tools %}\n{%- endif %}\n{%- if not tools_in_user_message is defined %}\n    {%- set tools_in_user_message = true %}\n{%- endif %}\n{%- if not date_string is defined %}\n    {%- if strftime_now is defined %}\n        {%- set date_string = strftime_now(\"%d %b %Y\") %}\n    {%- else %}\n        {%- set date_string = \"26 Jul 2024\" %}\n    {%- endif %}\n{%- endif %}\n{%- if not tools is defined %}\n    {%- set tools = none %}\n{%- endif %}\n\n{#- This block extracts the system message, so we can slot it into the right place. #}\n{%- if messages[0]['role'] == 'system' %}\n    {%- set system_message = messages[0]['content']|trim %}\n    {%- set messages = messages[1:] %}\n{%- else %}\n    {%- set system_message = \"\" %}\n{%- endif %}\n\n{#- System message #}\n{{- \"<|start_header_id|>system<|end_header_id|>\\n\\n\" }}\n{%- if tools is not none %}\n    {{- \"Environment: ipython\\n\" }}\n{%- endif %}\n{{- \"Cutting Knowledge Date: December 2023\\n\" }}\n{{- \"Today Date: \" + date_string + \"\\n\\n\" }}\n{%- if tools is not none and not tools_in_user_message %}\n    {{- \"You have access to the following functions. To call a function, please respond with JSON for a function call.\" }}\n    {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n    {{- \"Do not use variables.\\n\\n\" }}\n    {%- for t in tools %}\n        {{- t | tojson(indent=4) }}\n        {{- \"\\n\\n\" }}\n    {%- endfor %}\n{%- endif %}\n{{- system_message }}\n{{- \"<|eot_id|>\" }}\n\n{#- Custom tools are passed in a user message with some extra guidance #}\n{%- if tools_in_user_message and not tools is none %}\n    {#- Extract the first user message so we can plug it in here #}\n    {%- if messages | length != 0 %}\n        {%- set first_user_message = messages[0]['content']|trim %}\n        {%- set messages = messages[1:] %}\n    {%- else %}\n        {{- raise_exception(\"Cannot put tools in the first user message when there's no first user message!\") }}\n{%- endif %}\n    {{- '<|start_header_id|>user<|end_header_id|>\\n\\n' -}}\n    {{- \"Given the following functions, please respond with a JSON for a function call \" }}\n    {{- \"with its proper arguments that best answers the given prompt.\\n\\n\" }}\n    {{- 'Respond in the format {\"name\": function name, \"parameters\": dictionary of argument name and its value}.' }}\n    {{- \"Do not use variables.\\n\\n\" }}\n    {%- for t in tools %}\n        {{- t | tojson(indent=4) }}\n        {{- \"\\n\\n\" }}\n    {%- endfor %}\n    {{- first_user_message + \"<|eot_id|>\"}}\n{%- endif %}\n\n{%- for message in messages %}\n    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}\n        {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\\n\\n'+ message['content'] | trim + '<|eot_id|>' }}\n    {%- elif 'tool_calls' in message %}\n        {%- if not message.tool_calls|length == 1 %}\n            {{- raise_exception(\"This model only supports single tool-calls at once!\") }}\n        {%- endif %}\n        {%- set tool_call = message.tool_calls[0].function %}\n        {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' -}}\n        {{- '{\"name\": \"' + tool_call.name + '\", ' }}\n        {{- '\"parameters\": ' }}\n        {{- tool_call.arguments | tojson }}\n        {{- \"}\" }}\n        {{- \"<|eot_id|>\" }}\n    {%- elif message.role == \"tool\" or message.role == \"ipython\" %}\n        {{- \"<|start_header_id|>ipython<|end_header_id|>\\n\\n\" }}\n        {%- if message.content is mapping or message.content is iterable %}\n            {{- message.content | tojson }}\n        {%- else %}\n            {{- message.content }}\n        {%- endif %}\n        {{- \"<|eot_id|>\" }}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|start_header_id|>assistant<|end_header_id|>\\n\\n' }}\n{%- endif %}\n"
  lora_configs:
  - name: lora
    lora_rank: 32
    lora_alpha: 32
cluster:
  node_num: 1
  gpu_per_node: 8
buffer:
  batch_size: 96
  total_epochs: 1
  explorer_input:
    taskset:
      name: taskset
      storage_type: file
      path: openai/gsm8k
      split: train
      subset_name: main
      format:
        prompt_key: question
        response_key: answer
    default_workflow_type: math_workflow
  trainer_input:
    experience_buffer:
      name: experience_buffer
      storage_type: queue
explorer:
  runner_per_model: 16
  rollout_model:
    engine_num: 4
    tensor_parallel_size: 1
    gpu_memory_utilization: 0.9
    dtype: bfloat16
    seed: 42
trainer:
  trainer_type: verl
  save_interval: 100
  grad_clip: 1.0
monitor:
  monitor_type: wandb
synchronizer:
  sync_method: checkpoint
  sync_style: fixed
  sync_interval: 1
  sync_offset: 1
  sync_timeout: 1200

结果说明#

由于 Llama-3.2-3B 是基础（非指令微调）模型，其格式化指令跟随能力有限，且本实验仅训练了一个 epoch。因此，两种后端的最终 reward 都略高于 0.1。但训练曲线显示 reward 呈明显上升趋势，表明模型已成功学习。结果可视化如下：

GSM8K 训练奖励曲线

目录