trl/pr_5607/en/_app/immutable/nodes/19.44de3417.js

import{s as ke,n as Ue,o as xe}from"../chunks/scheduler.7b731bd4.js";import{S as Je,i as Ge,e as r,s as t,c as p,h as Ce,a as s,d as n,b as o,f as O,g as _,j as E,k as A,l as i,m as l,n as c,t as d,o as v,p as g}from"../chunks/index.cc268345.js";import{C as Pe,H as _e,E as Ie}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.f0d99f98.js";import{D as z}from"../chunks/Docstring.03f7b462.js";import{C as Be}from"../chunks/CodeBlock.169a125f.js";function Oe(be){let f,S,L,R,M,Y,N,D,w,ye='This feature implements the GFPO algorithm to enforce concise reasoning in the model’s output generation, as proposed in the paper <a href="https://huggingface.co/papers/2508.09726" rel="nofollow">Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning</a>.',q,T,K,F,Me="To activate GFPO in <code>GFPOTrainer</code>:",ee,$,Ne="<li>set <code>num_remains_in_group</code> in <code>GFPOConfig</code></li> <li>define a group filter function and set it to <code>group_filter_func</code> in <code>GFPOTrainer</code>. <code>group_filter_func</code> will score the <code>num_generations</code> completions and The GFPOTrainer filters groups according to their scores to get top <code>num_remains_in_group</code> completions as a new group. Model will be trained on the filtered group.</li>",ae,j,ne,k,te,m,U,ce,h,x,de,H,we="Main training entry point.",ve,u,J,ge,V,Te="Will save the model, so you can reload it using <code>from_pretrained()</code>.",ue,Z,Fe="Will only save from the main process.",fe,b,G,he,Q,$e="Upload <code>self.model</code> and <code>self.processing_class</code> to the 🤗 model hub on the repo <code>self.args.hub_model_id</code>.",oe,C,le,P,I,re,B,se,X,ie;return M=new Pe({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),N=new _e({props:{title:"GFPO",local:"gfpo",headingTag:"h1"}}),T=new _e({props:{title:"Usage",local:"usage",headingTag:"h2"}}),j=new Be({props:{code:"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",highlighted:`<span class="hljs-comment"># train_gfpo.py</span>
<span class="hljs-keyword">from</span> trl.experimental.gfpo <span class="hljs-keyword">import</span> GFPOConfig, GFPOTrainer

<span class="hljs-comment"># dummy group filter to scores the completions based on its indice in group</span>
<span class="hljs-keyword">class</span> <span class="hljs-title class_">GroupFilter</span>:
    <span class="hljs-keyword">def</span> <span class="hljs-title function_">__call__</span>(<span class="hljs-params">self, group_completions, group_rewards, **kwargs</span>):
        group_scores = []
        <span class="hljs-keyword">for</span> completions, rewards <span class="hljs-keyword">in</span> <span class="hljs-built_in">zip</span>(group_completions, group_rewards):
            scores = [<span class="hljs-built_in">float</span>(i) <span class="hljs-keyword">for</span> i <span class="hljs-keyword">in</span> <span class="hljs-built_in">range</span>(<span class="hljs-built_in">len</span>(completions))]
            group_scores.append(scores)
        <span class="hljs-keyword">return</span> group_scores

training_args = GFPOConfig(
    output_dir=<span class="hljs-string">&quot;Qwen3-0.6B-GFPO&quot;</span>,
    per_device_train_batch_size=<span class="hljs-number">4</span>,
    num_remains_in_group=<span class="hljs-number">2</span>,
    bf16=<span class="hljs-literal">True</span>,
)
trainer = GFPOTrainer(
    model=<span class="hljs-string">&quot;Qwen/Qwen3-0.6B&quot;</span>,
    reward_funcs=...,
    train_dataset=...,
    args=training_args,
    group_filter_func=GroupFilter(),
)
trainer.train()`,wrap:!1}}),k=new _e({props:{title:"GFPOTrainer",local:"trl.experimental.gfpo.GFPOTrainer",headingTag:"h2"}}),U=new z({props:{name:"class trl.experimental.gfpo.GFPOTrainer",anchor:"trl.experimental.gfpo.GFPOTrainer",parameters:[{name:"model",val:""},{name:"reward_funcs",val:""},{name:"args",val:" = None"},{name:"train_dataset",val:" = None"},{name:"eval_dataset",val:" = None"},{name:"processing_class",val:" = None"},{name:"reward_processing_classes",val:" = None"},{name:"group_filter_func",val:" = None"},{name:"callbacks",val:" = None"},{name:"optimizers",val:" = (None, None)"},{name:"peft_config",val:" = None"}],source:"https://github.com/huggingface/trl/blob/vr_5607/trl/experimental/gfpo/gfpo_trainer.py#L33"}}),x=new z({props:{name:"train",anchor:"trl.experimental.gfpo.GFPOTrainer.train",parameters:[{name:"resume_from_checkpoint",val:": str | bool | None = None"},{name:"trial",val:": optuna.Trial | dict[str, Any] | None = None"},{name:"ignore_keys_for_eval",val:": list[str] | None = None"}],parametersDescription:[{anchor:"trl.experimental.gfpo.GFPOTrainer.train.resume_from_checkpoint",description:`<strong>resume_from_checkpoint</strong> (<code>str</code> or <code>bool</code>, <em>optional</em>) &#x2014;
If a <code>str</code>, local path to a saved checkpoint as saved by a previous instance of <code>Trainer</code>. If a
<code>bool</code> and equals <code>True</code>, load the last checkpoint in <em>args.output_dir</em> as saved by a previous instance
of <code>Trainer</code>. If present, training will resume from the model/optimizer/scheduler states loaded here.`,name:"resume_from_checkpoint"},{anchor:"trl.experimental.gfpo.GFPOTrainer.train.trial",description:`<strong>trial</strong> (<code>optuna.Trial</code> or <code>dict[str, Any]</code>, <em>optional</em>) &#x2014;
The trial run or the hyperparameter dictionary for hyperparameter search.`,name:"trial"},{anchor:"trl.experimental.gfpo.GFPOTrainer.train.ignore_keys_for_eval",description:`<strong>ignore_keys_for_eval</strong> (<code>list[str]</code>, <em>optional</em>) &#x2014;
A list of keys in the output of your model (if it is a dictionary) that should be ignored when
gathering predictions for evaluation during the training.`,name:"ignore_keys_for_eval"}],source:"https://github.com/huggingface/trl/blob/vr_5607/transformers/trainer.py#L1323",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


<p>Object containing the global step count, training loss, and metrics.</p>
`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


<p><code>~trainer_utils.TrainOutput</code></p>
`}}),J=new z({props:{name:"save_model",anchor:"trl.experimental.gfpo.GFPOTrainer.save_model",parameters:[{name:"output_dir",val:": str | None = None"},{name:"_internal_call",val:": bool = False"}],source:"https://github.com/huggingface/trl/blob/vr_5607/transformers/trainer.py#L3746"}}),G=new z({props:{name:"push_to_hub",anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub",parameters:[{name:"commit_message",val:": str | None = 'End of training'"},{name:"blocking",val:": bool = True"},{name:"token",val:": str | None = None"},{name:"revision",val:": str | None = None"},{name:"**kwargs",val:""}],parametersDescription:[{anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub.commit_message",description:`<strong>commit_message</strong> (<code>str</code>, <em>optional</em>, defaults to <code>&quot;End of training&quot;</code>) &#x2014;
Message to commit while pushing.`,name:"commit_message"},{anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub.blocking",description:`<strong>blocking</strong> (<code>bool</code>, <em>optional</em>, defaults to <code>True</code>) &#x2014;
Whether the function should return only when the <code>git push</code> has finished.`,name:"blocking"},{anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub.token",description:`<strong>token</strong> (<code>str</code>, <em>optional</em>, defaults to <code>None</code>) &#x2014;
Token with write permission to overwrite Trainer&#x2019;s original args.`,name:"token"},{anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub.revision",description:`<strong>revision</strong> (<code>str</code>, <em>optional</em>) &#x2014;
The git revision to commit from. Defaults to the head of the &#x201C;main&#x201D; branch.`,name:"revision"},{anchor:"trl.experimental.gfpo.GFPOTrainer.push_to_hub.kwargs",description:`<strong>kwargs</strong> (<code>dict[str, Any]</code>, <em>optional</em>) &#x2014;
Additional keyword arguments passed along to <code>~Trainer.create_model_card</code>.`,name:"kwargs"}],source:"https://github.com/huggingface/trl/blob/vr_5607/transformers/trainer.py#L3993",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


<p>The URL of the repository where the model was pushed if <code>blocking=False</code>, or a <code>Future</code> object tracking the
progress of the commit if <code>blocking=True</code>.</p>
`}}),C=new _e({props:{title:"GFPOConfig",local:"trl.experimental.gfpo.GFPOConfig",headingTag:"h2"}}),I=new z({props:{name:"class trl.experimental.gfpo.GFPOConfig",anchor:"trl.experimental.gfpo.GFPOConfig",parameters:[{name:"output_dir",val:": str | None = None"},{name:"per_device_train_batch_size",val:": int = 8"},{name:"num_train_epochs",val:": float = 3.0"},{name:"max_steps",val:": int = -1"},{name:"learning_rate",val:": float = 1e-06"},{name:"lr_scheduler_type",val:": transformers.trainer_utils.SchedulerType | str = 'linear'"},{name:"lr_scheduler_kwargs",val:": dict | str | None = None"},{name:"warmup_steps",val:": float = 0"},{name:"optim",val:": transformers.training_args.OptimizerNames | str = 'adamw_torch_fused'"},{name:"optim_args",val:": str | None = None"},{name:"weight_decay",val:": float = 0.0"},{name:"adam_beta1",val:": float = 0.9"},{name:"adam_beta2",val:": float = 0.999"},{name:"adam_epsilon",val:": float = 1e-08"},{name:"optim_target_modules",val:": None | str | list[str] = None"},{name:"gradient_accumulation_steps",val:": int = 1"},{name:"average_tokens_across_devices",val:": bool = True"},{name:"max_grad_norm",val:": float = 1.0"},{name:"label_smoothing_factor",val:": float = 0.0"},{name:"bf16",val:": bool | None = None"},{name:"fp16",val:": bool = False"},{name:"bf16_full_eval",val:": bool = False"},{name:"fp16_full_eval",val:": bool = False"},{name:"tf32",val:": bool | None = None"},{name:"gradient_checkpointing",val:": bool = True"},{name:"gradient_checkpointing_kwargs",val:": dict[str, typing.Any] | str | None = None"},{name:"torch_compile",val:": bool = False"},{name:"torch_compile_backend",val:": str | None = None"},{name:"torch_compile_mode",val:": str | None = None"},{name:"use_liger_kernel",val:": bool = False"},{name:"liger_kernel_config",val:": dict[str, bool] | None = None"},{name:"use_cache",val:": bool = False"},{name:"neftune_noise_alpha",val:": float | None = None"},{name:"torch_empty_cache_steps",val:": int | None = None"},{name:"auto_find_batch_size",val:": bool = False"},{name:"logging_strategy",val:": transformers.trainer_utils.IntervalStrategy | str = 'steps'"},{name:"logging_steps",val:": float = 10"},{name:"logging_first_step",val:": bool = False"},{name:"log_on_each_node",val:": bool = True"},{name:"logging_nan_inf_filter",val:": bool = True"},{name:"include_num_input_tokens_seen",val:": str | bool = 'no'"},{name:"log_level",val:": str = 'passive'"},{name:"log_level_replica",val:": str = 'warning'"},{name:"disable_tqdm",val:": bool | None = None"},{name:"report_to",val:": None | str | list[str] = 'none'"},{name:"run_name",val:": str | None = None"},{name:"project",val:": str = 'huggingface'"},{name:"trackio_space_id",val:": str | None = 'trackio'"},{name:"eval_strategy",val:": transformers.trainer_utils.IntervalStrategy | str = 'no'"},{name:"eval_steps",val:": float | None = None"},{name:"eval_delay",val:": float = 0"},{name:"per_device_eval_batch_size",val:": int = 8"},{name:"prediction_loss_only",val:": bool = False"},{name:"eval_on_start",val:": bool = False"},{name:"eval_do_concat_batches",val:": bool = True"},{name:"eval_use_gather_object",val:": bool = False"},{name:"eval_accumulation_steps",val:": int | None = None"},{name:"include_for_metrics",val:": list = <factory>"},{name:"batch_eval_metrics",val:": bool = False"},{name:"save_only_model",val:": bool = False"},{name:"save_strategy",val:": transformers.trainer_utils.SaveStrategy | str = 'steps'"},{name:"save_steps",val:": float = 500"},{name:"save_on_each_node",val:": bool = False"},{name:"save_total_limit",val:": int | None = None"},{name:"enable_jit_checkpoint",val:": bool = False"},{name:"push_to_hub",val:": bool = False"},{name:"hub_token",val:": str | None = None"},{name:"hub_private_repo",val:": bool | None = None"},{name:"hub_model_id",val:": str | None = None"},{name:"hub_strategy",val:": transformers.trainer_utils.HubStrategy | str = 'every_save'"},{name:"hub_always_push",val:": bool = False"},{name:"hub_revision",val:": str | None = None"},{name:"load_best_model_at_end",val:": bool = False"},{name:"metric_for_best_model",val:": str | None = None"},{name:"greater_is_better",val:": bool | None = None"},{name:"ignore_data_skip",val:": bool = False"},{name:"restore_callback_states_from_checkpoint",val:": bool = False"},{name:"full_determinism",val:": bool = False"},{name:"seed",val:": int = 42"},{name:"data_seed",val:": int | None = None"},{name:"use_cpu",val:": bool = False"},{name:"accelerator_config",val:": dict | str | None = None"},{name:"parallelism_config",val:": accelerate.parallelism_config.ParallelismConfig | None = None"},{name:"dataloader_drop_last",val:": bool = False"},{name:"dataloader_num_workers",val:": int = 0"},{name:"dataloader_pin_memory",val:": bool = True"},{name:"dataloader_persistent_workers",val:": bool = False"},{name:"dataloader_prefetch_factor",val:": int | None = None"},{name:"remove_unused_columns",val:": bool | None = False"},{name:"label_names",val:": list[str] | None = None"},{name:"train_sampling_strategy",val:": str = 'random'"},{name:"length_column_name",val:": str = 'length'"},{name:"ddp_find_unused_parameters",val:": bool | None = None"},{name:"ddp_bucket_cap_mb",val:": int | None = None"},{name:"ddp_broadcast_buffers",val:": bool | None = None"},{name:"ddp_backend",val:": str | None = None"},{name:"ddp_timeout",val:": int = 1800"},{name:"fsdp",val:": list[transformers.trainer_utils.FSDPOption] | str | None = None"},{name:"fsdp_config",val:": dict[str, typing.Any] | str | None = None"},{name:"deepspeed",val:": dict | str | None = None"},{name:"debug",val:": str | list[transformers.debug_utils.DebugOption] = ''"},{name:"skip_memory_metrics",val:": bool = True"},{name:"do_train",val:": bool = False"},{name:"do_eval",val:": bool = False"},{name:"do_predict",val:": bool = False"},{name:"resume_from_checkpoint",val:": str | None = None"},{name:"warmup_ratio",val:": float | None = None"},{name:"logging_dir",val:": str | None = None"},{name:"local_rank",val:": int = -1"},{name:"model_init_kwargs",val:": dict[str, typing.Any] | str | None = None"},{name:"disable_dropout",val:": bool = False"},{name:"cast_lm_head_to_fp32",val:": bool = False"},{name:"num_generations",val:": int | None = 8"},{name:"num_generations_eval",val:": int | None = None"},{name:"max_completion_length",val:": int | None = 256"},{name:"ds3_gather_for_generation",val:": bool = True"},{name:"shuffle_dataset",val:": bool | None = True"},{name:"pad_to_multiple_of",val:": int | None = None"},{name:"generation_batch_size",val:": int | None = None"},{name:"steps_per_generation",val:": int | None = None"},{name:"temperature",val:": float = 1.0"},{name:"top_p",val:": float = 1.0"},{name:"top_k",val:": int = 0"},{name:"min_p",val:": float | None = None"},{name:"generation_kwargs",val:": dict | None = None"},{name:"chat_template_kwargs",val:": dict | None = None"},{name:"repetition_penalty",val:": float = 1.0"},{name:"cache_implementation",val:": str | None = None"},{name:"use_vllm",val:": bool = False"},{name:"vllm_mode",val:": str = 'colocate'"},{name:"vllm_model_impl",val:": str = 'vllm'"},{name:"vllm_enable_sleep_mode",val:": bool = False"},{name:"vllm_structured_outputs_regex",val:": str | None = None"},{name:"vllm_server_base_url",val:": str | None = None"},{name:"vllm_server_host",val:": str = '0.0.0.0'"},{name:"vllm_server_port",val:": int = 8000"},{name:"vllm_server_timeout",val:": float = 240.0"},{name:"vllm_group_port",val:": int = 51216"},{name:"vllm_gpu_memory_utilization",val:": float = 0.3"},{name:"vllm_max_model_length",val:": int | None = None"},{name:"vllm_tensor_parallel_size",val:": int = 1"},{name:"beta",val:": float = 0.0"},{name:"num_iterations",val:": int = 1"},{name:"epsilon",val:": float = 0.2"},{name:"delta",val:": float | None = None"},{name:"epsilon_high",val:": float | None = None"},{name:"sapo_temperature_neg",val:": float = 1.05"},{name:"sapo_temperature_pos",val:": float = 1.0"},{name:"vespo_k_pos",val:": float = 2.0"},{name:"vespo_lambda_pos",val:": float = 3.0"},{name:"vespo_k_neg",val:": float = 3.0"},{name:"vespo_lambda_neg",val:": float = 2.0"},{name:"importance_sampling_level",val:": str = 'token'"},{name:"reward_weights",val:": list[float] | None = None"},{name:"multi_objective_aggregation",val:": str = 'sum_then_normalize'"},{name:"scale_rewards",val:": str = 'group'"},{name:"loss_type",val:": str = 'dapo'"},{name:"mask_truncated_completions",val:": bool = False"},{name:"sync_ref_model",val:": bool = False"},{name:"ref_model_mixup_alpha",val:": float = 0.6"},{name:"ref_model_sync_steps",val:": int = 512"},{name:"top_entropy_quantile",val:": float = 1.0"},{name:"max_tool_calling_iterations",val:": int | None = None"},{name:"vllm_importance_sampling_correction",val:": bool = True"},{name:"vllm_importance_sampling_mode",val:": str = 'sequence_mask'"},{name:"vllm_importance_sampling_cap",val:": float = 3.0"},{name:"off_policy_mask_threshold",val:": float | None = None"},{name:"use_bias_correction_kl",val:": bool = False"},{name:"log_completions",val:": bool = False"},{name:"num_completions_to_print",val:": int | None = None"},{name:"log_unique_prompts",val:": bool = False"},{name:"log_completions_hub_repo",val:": str | None = None"},{name:"use_transformers_paged",val:": bool = False"},{name:"num_remains_in_group",val:": int | None = None"}],source:"https://github.com/huggingface/trl/blob/vr_5607/trl/experimental/gfpo/gfpo_config.py#L21"}}),B=new Ie({props:{source:"https://github.com/huggingface/trl/blob/main/docs/source/gfpo.md"}}),{c(){f=r("meta"),S=t(),L=r("p"),R=t(),p(M.$$.fragment),Y=t(),p(N.$$.fragment),D=t(),w=r("p"),w.innerHTML=ye,q=t(),p(T.$$.fragment),K=t(),F=r("p"),F.innerHTML=Me,ee=t(),$=r("ul"),$.innerHTML=Ne,ae=t(),p(j.$$.fragment),ne=t(),p(k.$$.fragment),te=t(),m=r("div"),p(U.$$.fragment),ce=t(),h=r("div"),p(x.$$.fragment),de=t(),H=r("p"),H.textContent=we,ve=t(),u=r("div"),p(J.$$.fragment),ge=t(),V=r("p"),V.innerHTML=Te,ue=t(),Z=r("p"),Z.textContent=Fe,fe=t(),b=r("div"),p(G.$$.fragment),he=t(),Q=r("p"),Q.innerHTML=$e,oe=t(),p(C.$$.fragment),le=t(),P=r("div"),p(I.$$.fragment),re=t(),p(B.$$.fragment),se=t(),X=r("p"),this.h()},l(e){const a=Ce("svelte-u9bgzb",document.head);f=s(a,"META",{name:!0,content:!0}),a.forEach(n),S=o(e),L=s(e,"P",{}),O(L).forEach(n),R=o(e),_(M.$$.fragment,e),Y=o(e),_(N.$$.fragment,e),D=o(e),w=s(e,"P",{"data-svelte-h":!0}),E(w)!=="svelte-1dy87lu"&&(w.innerHTML=ye),q=o(e),_(T.$$.fragment,e),K=o(e),F=s(e,"P",{"data-svelte-h":!0}),E(F)!=="svelte-1lfn8in"&&(F.innerHTML=Me),ee=o(e),$=s(e,"UL",{"data-svelte-h":!0}),E($)!=="svelte-1h9orgs"&&($.innerHTML=Ne),ae=o(e),_(j.$$.fragment,e),ne=o(e),_(k.$$.fragment,e),te=o(e),m=s(e,"DIV",{class:!0});var y=O(m);_(U.$$.fragment,y),ce=o(y),h=s(y,"DIV",{class:!0});var me=O(h);_(x.$$.fragment,me),de=o(me),H=s(me,"P",{"data-svelte-h":!0}),E(H)!=="svelte-1cilnet"&&(H.textContent=we),me.forEach(n),ve=o(y),u=s(y,"DIV",{class:!0});var W=O(u);_(J.$$.fragment,W),ge=o(W),V=s(W,"P",{"data-svelte-h":!0}),E(V)!=="svelte-r8h4ov"&&(V.innerHTML=Te),ue=o(W),Z=s(W,"P",{"data-svelte-h":!0}),E(Z)!=="svelte-1e6bius"&&(Z.textContent=Fe),W.forEach(n),fe=o(y),b=s(y,"DIV",{class:!0});var pe=O(b);_(G.$$.fragment,pe),he=o(pe),Q=s(pe,"P",{"data-svelte-h":!0}),E(Q)!=="svelte-8tudwd"&&(Q.innerHTML=$e),pe.forEach(n),y.forEach(n),oe=o(e),_(C.$$.fragment,e),le=o(e),P=s(e,"DIV",{class:!0});var je=O(P);_(I.$$.fragment,je),je.forEach(n),re=o(e),_(B.$$.fragment,e),se=o(e),X=s(e,"P",{}),O(X).forEach(n),this.h()},h(){A(f,"name","hf:doc:metadata"),A(f,"content",Ee),A(h,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),A(u,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),A(b,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),A(m,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),A(P,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8")},m(e,a){i(document.head,f),l(e,S,a),l(e,L,a),l(e,R,a),c(M,e,a),l(e,Y,a),c(N,e,a),l(e,D,a),l(e,w,a),l(e,q,a),c(T,e,a),l(e,K,a),l(e,F,a),l(e,ee,a),l(e,$,a),l(e,ae,a),c(j,e,a),l(e,ne,a),c(k,e,a),l(e,te,a),l(e,m,a),c(U,m,null),i(m,ce),i(m,h),c(x,h,null),i(h,de),i(h,H),i(m,ve),i(m,u),c(J,u,null),i(u,ge),i(u,V),i(u,ue),i(u,Z),i(m,fe),i(m,b),c(G,b,null),i(b,he),i(b,Q),l(e,oe,a),c(C,e,a),l(e,le,a),l(e,P,a),c(I,P,null),l(e,re,a),c(B,e,a),l(e,se,a),l(e,X,a),ie=!0},p:Ue,i(e){ie||(d(M.$$.fragment,e),d(N.$$.fragment,e),d(T.$$.fragment,e),d(j.$$.fragment,e),d(k.$$.fragment,e),d(U.$$.fragment,e),d(x.$$.fragment,e),d(J.$$.fragment,e),d(G.$$.fragment,e),d(C.$$.fragment,e),d(I.$$.fragment,e),d(B.$$.fragment,e),ie=!0)},o(e){v(M.$$.fragment,e),v(N.$$.fragment,e),v(T.$$.fragment,e),v(j.$$.fragment,e),v(k.$$.fragment,e),v(U.$$.fragment,e),v(x.$$.fragment,e),v(J.$$.fragment,e),v(G.$$.fragment,e),v(C.$$.fragment,e),v(I.$$.fragment,e),v(B.$$.fragment,e),ie=!1},d(e){e&&(n(S),n(L),n(R),n(Y),n(D),n(w),n(q),n(K),n(F),n(ee),n($),n(ae),n(ne),n(te),n(m),n(oe),n(le),n(P),n(re),n(se),n(X)),n(f),g(M,e),g(N,e),g(T,e),g(j,e),g(k,e),g(U),g(x),g(J),g(G),g(C,e),g(I),g(B,e)}}}const Ee='{"title":"GFPO","local":"gfpo","sections":[{"title":"Usage","local":"usage","sections":[],"depth":2},{"title":"GFPOTrainer","local":"trl.experimental.gfpo.GFPOTrainer","sections":[],"depth":2},{"title":"GFPOConfig","local":"trl.experimental.gfpo.GFPOConfig","sections":[],"depth":2}],"depth":1}';function Ae(be){return xe(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class Le extends Je{constructor(f){super(),Ge(this,f,Ae,Oe,ke,{})}}export{Le as component};