jndZddlmZddlZddlZddlZddlZddlZddlZddl m Z ddl m Z ddl m Z ddlZddlZddlmZmZmZddlmZmZmZmZmZmZdd lmZdd lmZdd l m!Z!Gd d Z"dS)a5 Module provides functionalities for hyperparameter tuning of the Ultralytics YOLO models for object detection, instance segmentation, image classification, pose estimation, and multi-object tracking. Hyperparameter tuning is the process of systematically searching for the optimal set of hyperparameters that yield the best model performance. This is particularly crucial in deep learning models like YOLO, where small changes in hyperparameters can lead to significant differences in model accuracy and efficiency. Examples: Tune hyperparameters for YOLO26n on COCO8 at imgsz=640 and epochs=10 for 300 tuning iterations. >>> from ultralytics import YOLO >>> model = YOLO("yolo26n.pt") >>> model.tune(data="coco8.yaml", epochs=10, iterations=300, optimizer="AdamW", plots=False, save=False, val=False) ) annotationsN)Counter)datetime)Path) CFG_INT_KEYSget_cfg get_save_dir) DEFAULT_CFGLOGGERYAML callbackscolorstrremove_colorstr)check_requirements) torch_load)plot_tune_resultsceZdZdZedfdCdZdDdEd ZdFd ZdGdHdZe dZ dIdJdZ dKd Z d!Z dLd#ZdIdMd'ZdNd)Ze dOd*Ze dPdQd.ZedRd0Ze dSd3Ze dTdUd9Z dVdWd=ZdXdYdBZdS)ZTuneraSA class for hyperparameter tuning of YOLO models. The class evolves YOLO model hyperparameters over a given number of iterations by mutating them according to the search space and retraining the model to evaluate their performance. Supports both local NDJSON storage and distributed MongoDB Atlas coordination for multi-machine hyperparameter optimization. Attributes: space (dict[str, tuple]): Hyperparameter search space containing bounds and scaling factors for mutation. tune_dir (Path): Directory where evolution logs and results will be saved. tune_file (Path): Path to the NDJSON file where evolution logs are saved. args (SimpleNamespace): Configuration arguments for the tuning process. callbacks (dict): Callback functions to be executed during tuning. prefix (str): Prefix string for logging messages. mongodb (MongoClient): Optional MongoDB client for distributed tuning. collection (Collection): MongoDB collection for storing tuning results. Methods: _mutate: Mutate hyperparameters based on bounds and scaling factors. __call__: Execute the hyperparameter evolution across multiple iterations. Examples: Tune hyperparameters for YOLO26n on COCO8 at imgsz=640 and epochs=10 for 300 tuning iterations. >>> from ultralytics import YOLO >>> model = YOLO("yolo26n.pt") >>> model.tune( >>> data="coco8.yaml", >>> epochs=10, >>> iterations=300, >>> plots=False, >>> save=False, >>> val=False >>> ) Tune with distributed MongoDB Atlas coordination across multiple machines: >>> model.tune( >>> data="coco8.yaml", >>> epochs=10, >>> iterations=300, >>> mongodb_uri="mongodb+srv://user:pass@cluster.mongodb.net/", >>> mongodb_db="ultralytics", >>> mongodb_collection="tune_results" >>> ) Tune with custom search space: >>> model.tune(space={"lr0": (1e-5, 1e-2), "momentum": (0.7, 0.98)}) N _callbacks dict | Nonec h|ddp@iddddddd d d d d dddddddddddddddddddddd d!d dddddddd!d" |_|d#d}|d$d%}|d&d'}t|(|_|jj|j_t |j|jjpd)*|_d+\|j_|j_|j_|jd,z |_ |ptj |_ td-|_ tj|d|_|r||||t#j|j d.|jd/|j d0dS)1zInitialize the Tuner with configurations. Args: args (dict): Configuration for hyperparameter evolution. _callbacks (dict | None, optional): Callback functions to be executed during tuning. spaceNlr0)gh㈵>{Gz?lrf)r?momentum)gffffff?g\(\?g333333? weight_decay)gMbP? warmup_epochs)rg@warmup_momentum)rgffffff?box)rg4@cls)皙?@cls_pw)rrdfl)g?g(@hsv_h)rr$hsv_s)rg?hsv_vdegrees)rgF@ translatescaleshear)rg$@) perspectiveflipudfliplrbgrmosaicmixupcutmix copy_paste close_mosaic mongodb_uri mongodb_db ultralyticsmongodb_collection tuner_results) overridestunename)NFFztune_results.ndjsonzTuner: z*Initialized Tuner instance with 'tune_dir=z' uT💡 Learn about tuning at https://docs.ultralytics.com/guides/hyperparameter-tuning)poprrargsresumeexist_okr r@tune_dir tune_filer get_default_callbacksrprefixadd_integration_callbacksmongodb _init_mongodbr info)selfrBrr8r9r;s [/var/www/html/Carbon-Document/venv/lib/python3.11/site-packages/ultralytics/engine/tuner.py__init__zTuner.__init__WsvXXgt,, 1 <1  ;1  ( 1 L 1 Z 1  {1  ;1  :1  j1  ;1  Z1  Z1  Z1  {1 !1 " [#1 $ [%1 &(    $'71 1 1  :hh}d33 XXlM:: !XX&:OLLd+++ !Y- $TYTY^5MvNNN ?S<  *DI,<)>>#Hy'F'H'Hy)) +D111  L   {J8J K K K { q qdm q q{ q q q     uristr max_retriesintc tdddlm}ddlm}m}t |D]} ||dddddd d d }|jd tj |j d|dzd|cS#||f$rK||dz krd|z}tj |j d|dzd|dtj|YwxYwdS)afCreate MongoDB client with exponential backoff retry on connection failures. Args: uri (str): MongoDB connection string with credentials and cluster information. max_retries (int): Maximum number of connection attempts before giving up. Returns: (MongoClient): Connected MongoDB client instance. pymongor) MongoClient)ConnectionFailureServerSelectionTimeoutErrori0ui Ni@TrRi`)serverSelectionTimeoutMSconnectTimeoutMSsocketTimeoutMS retryWrites retryReads maxPoolSize minPoolSize maxIdleTimeMSpingz$Connected to MongoDB Atlas (attempt )z#MongoDB connection failed (attempt z), retrying in zs...N)rrXrYpymongo.errorsrZr[rangeadmincommandr rLrHwarningtimesleep) rMrSrUrYrZr[attemptclient wait_times rN_connectzTuner._connectsj 9%%%''''''QQQQQQQQ[)) & &G &$-2%*$) $# " !"'    $$V,,, t{^^PWZ[P[^^^___ %'BC & & &kAo--wJ {rrwQR{rrclrrr 9%%%%% &! & &sABACCc|||_|j|||_|jdgdt j|jddS)aInitialize MongoDB connection for distributed tuning. Connects to MongoDB Atlas for distributed hyperparameter optimization across multiple machines. Each worker saves results to a shared collection and reads the latest best hyperparameters from all workers for evolution. Args: mongodb_uri (str): MongoDB connection string. mongodb_db (str, optional): Database name. mongodb_collection (str, optional): Collection name. Notes: - Creates a fitness index for fast queries of top results - Falls back to local NDJSON mode if connection fails - Uses connection pooling and retry logic for production reliability )fitnessT) backgroundz*Using MongoDB Atlas for distributed tuningN)rsrJ collection create_indexr rLrH)rMr8r9r;s rNrKzTuner._init_mongodbsj }}[11 ,z23EF $$o%64$HHH t{NNNOOOOOrPnreturnlistc t|jdd|S#t $rgcYSwxYw)zGet top N results from MongoDB sorted by fitness. Args: n (int): Number of top results to retrieve. Returns: (list[dict]): List of result documents with fitness scores and hyperparameters. rurv)r}rxfindsortlimit Exception)rMr{s rN_get_mongodb_resultszTuner._get_mongodb_resultsse ,,..33IrBBHHKKLL L   III sA A AAcht|dr|nt|S)z2Convert tensor-like values for JSON serialization.item)hasattrrrT)xs rN _json_defaultzTuner._json_defaults+#1f--9qvvxxx3q669rP iterationrufloathyperparametersdict[str, float]datasetsdict[str, dict] save_dirsdict[str, str] | Nonedictc>|t|d||d}|r||d<|S)z%Build one local tuning result record.rz)rrurrr)round)rMrrurrrresults rN_result_recordzTuner._result_records?#Wa((.     ,"+F;  rPmetricsc  |j|d|D||tj|ddS#t $r)}t j|jd|Yd}~dSd}~wwxYw)aSave results to MongoDB with proper type conversion. Args: fitness (float): Fitness score achieved with these hyperparameters. hyperparameters (dict[str, float]): Dictionary of hyperparameter values. metrics (dict): Complete training metrics dictionary (mAP, precision, recall, losses, etc.). datasets (dict[str, dict]): Per-dataset metrics for the iteration. iteration (int): Current iteration number. cbi|],\}}|t|dr|n|-S)r)rr.0kvs rN z*Tuner._save_to_mongodb.. s='w'w'wW[WXZ[68J8J,QAFFHHHPQ'w'w'wrP)rurrr timestamprzMongoDB save failed: N) rx insert_oneitemsrnowrr rmrH)rMrurrrres rN_save_to_mongodbzTuner._save_to_mongodbs" E O & &&'w'w_n_t_t_v_v'w'w'w& (!)!*       E E E NdkCCCC D D D D D D D D D EsAA BBBc~ t|jdd}|sdSt |jdd5}|D]}|tj| |d|dpd| d i| d i| d |j d z ddddS#1swxYwYdS#t$r)}tj|jd|Yd}~dSd}~wwxYw)aSync MongoDB results to the local NDJSON tuning log. Downloads all results from MongoDB and writes them to the local NDJSON file in chronological order. This keeps resume, mutation, and plotting on the same local source of truth when using distributed tuning. rrfNwutf-8encodingrurrrrdefault zMongoDB to NDJSON sync failed: )r}rxrropenrFwritejsondumpsrgetrrr rmrH)rM all_resultsfrrs rN_sync_mongodb_to_filezTuner._sync_mongodb_to_files  Ot3355::;JJKKK dncG<<< )  FGG  // &{ 3 &y 1 8S & +z-Tuner._load_local_results..7s/CCCdjjllCDJt$$CCCrPN)rFexistsr)rMrs rN_load_local_resultszTuner._load_local_results2s~$$&& I $.7 3 3 3 DqCCCCC D D D D D D D D D D D D D D D D D Ds A  AAresults int | Nonenp.ndarray | Nonec|sdStjfd|Dt}||Stj|dddf }||d|S)zKConvert local NDJSON records to a fitness-plus-hyperparameters numpy array.Ncjg|].ddgfdjDz/S)rurc g|]?}di|tj|@Sr)rgetattrrBrrrrMs rNrz..@sEbbbRS155*B//33Awty!7L7LMMbbbrP)rrrrrMs @rNrz1Tuner._local_results_to_array..>sa   y#&&'bbbbbW[Wabbbc   rP)dtyper)nparrayrargsort)rMrr{rorders` rN_local_results_to_arrayzTuner._local_results_to_array9s 4 H    !         9H AaaadG8$$x|rPrct|jdd5}|tj||jdzddddS#1swxYwYdS)z1Append one tuning result to the local NDJSON log.arrrrN)rrFrrrr)rMrrs rN_save_local_resultzTuner._save_local_resultJs $.# 8 8 8 KA GGDJvt/ABBBTI J J J K K K K K K K K K K K K K K K K K Ks2AAAc|di}t|dkr.tt|St|dkrd|DSdS)z*Summarize best-result metrics for logging.rrfc`i|]+\}}|t|dpdd,S)rurrz)rrrs rNrz'Tuner._best_metrics..Vs9VVVTQAuQUU9--4a88VVVrPN)rlennextitervaluesr)rrs rN _best_metricszTuner._best_metricsOs|::j"-- x==A  X__..//00 0 x==1  VVX^^EUEUVVV VtrPF require_allboolc|di}t|oC|r!t|n t |S)z9Return whether a tuning result contains training metrics.r)rrallrany)rrrs rN_has_training_metricszTuner._has_training_metricsYsY::j"--H~~e[#d3x'8'8#9#9#9cRZRaRaRcRcNdNderP np.ndarraycfdt|D}|r-|t||n t|S)zDReturn the best result index, preferring rows with training metrics.cDg|]\}}||Sr)r)rirr#s rNrz,Tuner._best_result_index..bs1\\\yq&#:S:STZ:[:[\\\\rP) enumeraterVargmax)r#rruvalids` rN_best_result_indexzTuner._best_result_index_sf]\\\Ig$6$6\\\6;VuS..001122W^^EUEUAVAVVrPdata list[str]cd|D}t|t}}g}|D]@}||xxdz cc<|||dkr |d||n|A|S)zGCreate stable unique dataset names for logging and per-run directories.cPg|]#}tt|j$Sr)rrTstem)rds rNrz(Tuner._dataset_names..hs'111qc!ff"111rPrf-)rappend)rstemstotalsseennamesrs rN_dataset_nameszTuner._dataset_nameses21D111u~~wyy O OD JJJ!OJJJ LL6$)weightsrc0g|]}|ddS)rfNr)rrrs rNrz$Tuner._crossover..ys%777Q!QRR777rPrr$)minrrisfinitersum ones_likerandomchoicesrjstackmaxwhereuniformshape) rrrridxs parents_matlohispans ` rN _crossoverzTuner._crossoverpsO 3q66NNAAAqD'AaaadGKKMM)D0{7##'')) ,W[[]]a-?-?l7++G~eCFFmmWBBBh7777$777;; ##[__Q%7%7BBwx 29#4#4T3 #K#KTRRy  edl!2B4EFFFrP?mutationsigmac d}jr|x}r!tjfd|D}nfjjjjvr..s?)s)s)s\]!,=*>*B*B1dimmTUFVFV*W*W)s)s)srP)rkeysrs @rNrz!Tuner._mutate..s]9)s)s)s)s)saeakapaparar)s)s)ssrPrc:g|]}tj|SrrrBrrrMs rNrz!Tuner._mutate..s%&X&X&Xwty!'<'<&X&X&XrP)r{cHg|]}t|dkr|dnd S)rRrhr)r)rrs rNrz!Tuner._mutate..s.XXXqc!ffkkaddsXXXrPrfrg?r%cTi|]$\}}|t||z%Sr)r)rrrfactorsgeness rNrz!Tuner._mutate..s4\\\tq!1eE!Hwqz122\\\rPc<i|]}|tj|Srrrs rNrz!Tuner._mutate..s'GGG1gdi++GGGrPrrzr7epochs)rJrrrrxr@databaselist_collection_namesrrrrr default_rngrr ronesrstandard_normalrexpcliprrrrr)rMr{r r rrrngnggainsmaskstephyprboundsrrs` @@rN_mutatez Tuner._mutates  < [33A666w [H!( %)A)W)W)Y)YYYHse&X&X&X&XdjooFWFW&X&X&XXYZZ 9,,T-E-E-G-G1,MMA =)''))CTZBOOA&&EHXXDJDUDUDWDWXXXYYEgbkkG&A&& Lzz"~~0**2..%%-@(4s;;@@sKK&A&& L]\\\\yIZIZ?[?[\\\CCGGGGTZ__5F5FGGGC))++ F FIAv3s3q66!955vayAA1EECFF S "'N(;"<"?|| dd fD} tjA|jd;z | t|,C|j d<dz=tjD|jd;z |r%tj |j d>|d?|d@d S d S)AaExecute the hyperparameter evolution process when the Tuner instance is called. This method iterates through the specified number of iterations, performing the following steps: 1. Sync MongoDB results to local NDJSON (if using distributed mode) 2. Mutate hyperparameters using the best previous results or defaults 3. Train a YOLO model with the mutated hyperparameters 4. Log fitness scores and hyperparameters to MongoDB and/or NDJSON 5. Track the best performing configuration across all iterations Args: iterations (int): The number of generations to run the evolution for. cleanup (bool): Whether to delete iteration weights to reduce storage space during tuning. T)parentsrDrrzResuming tuning run z from iteration rfz...gr@rrr$)r zStarting iteration /z with hyperparameters: rcLg|] }tt|!S)r?)r r)rr@ train_argss rNrz"Tuner.__call__..s.]]]tl7:#6#6TBBB]]]rPcg|]}|dz S)rr)rss rNrz"Tuner.__call__..s;;;Q1y=;;;rPsave_dirsysz-mzultralytics.cfg.__init__trainc3*K|]\}}|d|VdS)=Nrrs rN z!Tuner.__call__..s0.Y.Yda!zzazz.Y.Y.Y.Y.Y.YrP)checkzbest.ptzlast.pt train_metricsz5training failure for hyperparameter tuning iteration rNrurc4i|]\}}|t|Sr)rT)rdatasetr3s rNrz"Tuner.__call__..s$PPPZWa#a&&PPPrP)rFr) ignore_errorsz*.ptu complete ✅ucomplete (all failed) ❌z complete (u succeeded) ⚠️z iterations z (z.2fzs)zResults saved to boldz Best fitness=z observed at iteration zBest fitness metrics are zBest fitness model is rz"not saved for multi-dataset tuningzKNo iterations produced training metrics; skipping best_hyperparameters.yamlcdi|]-\}}||tvrt|nt|.Sr)rrVrrs rNrz"Tuner.__call__.._sEEIQAl!2!2s1vvvarPzbest_hyperparameters.yamlz# )rheaderzTarget iterations (z) reached in MongoDB (z ). Stopping.)ErnrEmkdirrJrrFrrrr rLrHrjrr)varsrBrA isinstancer}tuplerr rrziprT __import__ executabler subprocessrunrrogccollecttorchcuda empty_cachererrorrrrrrrrxcount_documentsrrrrshutilrmtreeglobcopy2rrextendrjoinrrr saverreplaceprint).rMr+r,t0best_save_dirs n_successfulstartrfracsigma_i mutated_hypr dataset_namesr4 weights_dirr all_fitnessdataset_metricsjrr= metrics_ilaunchcmd ckpt_filerrurstop_after_iterationtotal_mongo_iterationsrrbest_idx best_result n_attemptedcurrent_best_save_dirsbest_is_currentr3 weight_dirbest_weights_dirckptstatushas_valid_best header_linesrAr1s. @rN__call__zTuner.__call__s Y[[ D4888  "))$)FFF  < )  & & ( ( ( > " " k002233E K4;iiDMii[`cd[diii j j juj))N N Aq5y#&&DC$J&G,,W,55K K4;ss1q5ss:ssfqss t t t;DOO;{;J>>&))DdT5M22 v //55Mt99>>gj112233]]]]}]]]  <;(;;;KGK O#,S}-E-E#F#F D D>$+t+L+LMMPTT  4=+FFGGG#  {}}z}}Yo}}}  UN N sC4L ML;;M)rr)rQrR)rSrTrUrV)rQrQrQ)rz)r{rVr|r})N) rrVrurrrrrrrr|r) rurrrrrrrrrV)r|r)rrr{rr|r)rr)rrr|r)F)rrrrr|r)rrrurr|rV)rr}r|r)rr)rrrrrrVr|r)rr r)r{rVr rr rr|r)r*T)r+rVr,r)__name__ __module__ __qualname____doc__r rOrsrKr staticmethodrrrrrrrrr classmethodrrr r)rxrrPrNrr's9--^(49 9 9 9 9 v&&&&&&&&&&PPPPP*     ::\:,0 &EEEE>OOO<DDDD"KKKK \ffff\f WWW[W \ G G G G\ G" @@@@@DjjjjjjjrPr)#r| __future__rrKrrrRrIrn collectionsrrpathlibrnumpyrrMultralytics.cfgrrr ultralytics.utilsr r r r rrultralytics.utils.checksrultralytics.utils.patchesrultralytics.utils.plottingrrrrPrNrsg  #"""""   ??????????]]]]]]]]]]]]]]]]777777000000888888E E E E E E E E E E rP