U ܟj؉@sPUddlmZddlZddlZddlmZmZmZmZm Z ddl m Z ddl m ZGddde Zed d d Zd dd ddZd dddddZddgfd dddddZd^d dddddZddgfd dddddZd dd dd Zd dd d!d"Zd dd d#d$Zd dd d%d&Zd dd d'd(Zd dd d)d*Zd d+d,d-d.d/Zddgfddd0d1d2d3Zd dd d4d5Zd dd d6d7d8Zdd9d9d:d;d<Z d dd9d=d>d?d@Z!d dAd dBdCZ"dd9ddDdEdFZ#d dGd dHdIZ$d dd9d9dAdJdKdLZ%d dd dMdNZ&d_d+dddPdQdRZ'dSZ(dTe)dU<dVdWdXdYZ*d dd,dVd9dZdAd[d\d]Z+dS)`) annotationsN)AnyTypeVarCallableOptional NamedTuple) TypeAlias)pandasc@s^eZdZUded<dZded<dZded<dZded<dZded <dZded <dZ ded <dS) RemediationstrnameNz Optional[str] immediate_msg necessary_msgzOptional[Callable[[Any], Any]] necessary_fn optional_msg optional_fn error_msg) __name__ __module__ __qualname____annotations__rrrrrrrr:/tmp/pip-unpacked-wheel-2qj3ofc_/openai/lib/_validators.pyr s      r OptionalDataFrameTzOptional[pd.DataFrame])bound pd.DataFramedfreturncCs8d}t||krdnd}dt|d|}td|dS)z This validator will only print out the number of examples and recommend to the user to increase the number of examples if less than 100. dz. In general, we recommend having at least a few hundred examples. We've found that performance tends to linearly increase for every doubling of the number of examplesz - Your file contains z prompt-completion pairs num_examplesr r)lenr )rZ MIN_EXAMPLESZoptional_suggestionrrrrnum_examples_validators r%r )rnecessary_columnrcsddddddd}d}d}d}|jkrdd|jDkrtddd fd d }|}d d }dd}n dd}td||||dS)z[ This validator will ensure that the necessary column is present in the dataframe. rrrcolumnrcs2fdd|jD}|j|didd|S)Ncs g|]}t|kr|qSrr lower.0cr(rr -szInecessary_column_validator..lower_case_column..rT)columnsZinplace)r0renamer*)rr(colsrr.rlower_case_column,sz5necessary_column_validator..lower_case_columnNcSsg|]}t|qSrr)r+rrrr/7sz.necessary_column_validator..rcs |SNr)rr3r&rrlower_case_column_creator9sz=necessary_column_validator..lower_case_column_creatorz - The `z ` column/key should be lowercasezLower case column name to ``z^` column/key is missing. Please make sure you name your columns/keys appropriately, then retryr&)r rrrr)r0r )rr&rrrrr6rr5rnecessary_column_validator's&   r8prompt completionz list[str])rfieldsrcsg}d}d}d}t|jdkrfdd|jD}d}|D]8fdd|D}t|dkr:|dd d 7}q:d ||}d |}d d dfdd }td|||dS)zK This validator will remove additional columns from the dataframe. Nr csg|]}|kr|qSrrr+r;rrr/Usz/additional_column_validator..r!csg|]}|kr|qSrrr+)acrrr/Xsrz9 WARNING: Some of the additional columns/keys contain `z<` in their name. These will be ignored, and the column/key `z`` will be used instead. This could also result from a duplicate column/key in the provided file.zh - The input file should contain exactly two columns/keys per row. Additional columns/keys present are: z Remove additional columns/keys: rxrcs|Sr4rr?r<rrr^sz1additional_column_validator..necessary_fnZadditional_columnr rrr)r$r0r )rr;Zadditional_columnsrrr warn_messageZdupsr)r=r;radditional_column_validatorKs(  rC)rfieldrcsd}d}d}|dds2|r|dk|B}|j|}dd|}dddfd d }d t|d d }td|||dS)zA This validator will ensure that no completion is empty. NcSs|dkS)Nr!rr@rrrqz+non_empty_field_validator..r!z - `z?` column/key should not contain empty strings. These are rows: rr>cs||dkjgdS)Nr!Zsubset)Zdropnar@rDrrrvsz/non_empty_field_validator..necessary_fnRemove z rows with empty sZempty_rA)applyanyZisnull reset_indexindextolistr$r )rrDrrrZ empty_rowsZ empty_indexesrrHrnon_empty_field_validatoris&rPcs|jd}|j|}d}d}d}t|dkrzdt|ddd|}dt|d }d d d fd d }td|||dS)zY This validator will suggest to the user to remove duplicate rows if they exist. rGNr - There are z duplicated -z sets. These are rows: rIz duplicate rowsrr>cs |jdS)NrG)Zdrop_duplicatesr@r<rrrsz.duplicated_rows_validator..optional_fnduplicated_rowsr rrr)Z duplicatedrMrNrOr$joinr )rr;rSZduplicated_indexesrrrrr<rduplicated_rows_validators   rVcsd}d}d}t|}|dkr|ddddd|tdkr|d td d }d td }dddfdd }td|||dS)zW This validator will suggest to the user to remove examples that are too long. Nopen-ended generationrr)drcSs$|jdddd}|j|S)NcSst|jt|jdkS)Ni')r$r9r:r@rrrrErFzClong_examples_validator..get_long_indexes..)Zaxis)rKrMrNrO)rX long_examplesrrrget_long_indexessz1long_examples_validator..get_long_indexesrrQz. examples that are very long. These are rows: zf For conditional generation, and for classification the examples shouldn't be longer than 2048 tokens.rIz long examplesr>cs8|}|kr.tjdt|d|d||S)NzeThe indices of the long examples has changed as a result of a previously applied recommendation. The z? long examples to be dropped are now at the following indices:  )sysstdoutwriter$drop)r?Zlong_indexes_to_dropr[Z long_indexesrrrs z,long_examples_validator..optional_fnrZrT)infer_task_typer$r )rrrrft_typerrarlong_examples_validators" rdc sd}d}d}d}dddddddg}|D]@}|dkrH|jjd rHq(|jjj|d d r`q(|qjq(d d }t|}|d krtddSddddddt|jdd} |j| krd| d}td|dS| dkrP| d d } d| d}t | dkr|d|d7}|jjdt |  jj| d d rT|d| d7}nd }| dkrd!|d"}ddd#fd$d% }td&||||d'S)(z This validator will suggest to add a common suffix to the prompt if one doesn't already exist in case of classification or conditional generation. Nz ### => z ->z ### z === z --- z ===> z ---> r\Fregex\nrW common_suffixr rr?suffixrcSs|d|7<|SNr9rr?rkrrr add_suffixsz2common_prompt_suffix_validator..add_suffixrkxfixzAll prompts are identical: `zt` Consider leaving the prompts blank if you want to do open-ended generation, otherwise ensure prompts are differentr rr!z - All prompts end with suffix `r7 R. This suffix seems very long. Consider replacing with a shorter suffix, such as `z5 WARNING: Some of your prompts contain the suffix `zZ` more than once. We strongly suggest that you review your prompts and add a unique suffixa - Your data does not contain a common separator at the end of your prompts. Having a separator string appended to the end of the prompt makes it clearer to the fine-tuned model where the completion should begin. See https://platform.openai.com/docs/guides/fine-tuning/preparing-your-dataset for more detail and examples. If you intend to do open-ended generation, then you should leave the prompts emptyzAdd a suffix separator `z` to all promptsr>cs |Sr4rr@rnZsuggested_suffixrrrsz3common_prompt_suffix_validator..optional_fncommon_completion_suffixr rrrr) r9r containsrLreplacerbr get_common_xfixallr$) rrrrrsuffix_options suffix_optiondisplay_suggested_suffixrcrhcommon_suffix_new_line_handledrrtrcommon_prompt_suffix_validators^       (  rcsd}d}d}d}t|jdddkr0tddSdddd d d |jkrZtddSdkrd d }|tkr|d7}dd}dddfdd }td|||dS)zd This validator will suggest to remove a common prefix from the prompt if a long one exist. Nprefixror! common_prefixrir)r?rrcSs|djt|d|d<|Srlr r$)r?rrrrremove_common_prefixsz.remove_common_prefixz" - All prompts start with prefix `r7z. Fine-tuning doesn't require the instruction specifying the task, or a few-shot example scenario. Most of the time you should only add the input data into the prompt, and the desired output into the completionRemove prefix `z` from all promptsr>cs |Sr4rr@rrrrr!sz3common_prompt_prefix_validator..optional_fnZcommon_prompt_prefixrT)ryr9r rzr$rZMAX_PREFIX_LENrrrrrrcommon_prompt_prefix_validators,     rcsd}t|jddtdko(ddkt|kr@tddSddddd d d |jkrltddSd d }dd}dddfdd }td|||dS)zh This validator will suggest to remove a common prefix from the completion if a long one exist. rror rrir)r?r ws_prefixrcSs4|djt|d|d<|r0d|d|d<|S)Nr:rr)r?rrrrrr7sz@common_completion_prefix_validator..remove_common_prefixz& - All completions start with prefix `z_`. Most of the time you should only add the output data into the completion, without any prefixrz` from all completionsr>cs |Sr4rr@rrrrrrEsz7common_completion_prefix_validator..optional_fnZcommon_completion_prefixrT)ryr:r$r rzrrrr"common_completion_prefix_validator,s"     rc sd}d}d}d}t|}|dks(|dkr2tddSt|jdd}|j|krld|d |d }td|d Sd d ddddddddg }|D]$}|jjj|ddrq|qqd d} dddddd|dkrH|d d} d| d }t |dkr|d | d 7}|jjdt | jj|ddrL|d!|d"7}nd#}|dkrxd$| d%}ddd&fd'd( }td)||||d*S)+z This validator will suggest to add a common suffix to the completion if one doesn't already exist in case of classification or conditional generation. NrWclassificationrhrirkroz All completions are identical: `zJ` Ensure completions are different, otherwise the model will just repeat `r7rqz [END]r\.z ENDz***z+++z&&&z$$$z@@@z%%%FrergrrjcSs|d|7<|SNr:rrmrrrrnvsz6common_completion_suffix_validator..add_suffixr!z$ - All completions end with suffix `rrrsz9 WARNING: Some of your completions contain the suffix `zU` more than once. We suggest that you review your completions and add a unique endingaH - Your data does not contain a common ending at the end of your completions. Having a common ending string appended to the end of the completion makes it clearer to the fine-tuned model where the completion should end. See https://platform.openai.com/docs/guides/fine-tuning/preparing-your-dataset for more detail and examples.zAdd a suffix ending `z` to all completionsr>cs |Sr4rr@rtrrrsz7common_completion_suffix_validator..optional_fnrurv) rbr ryr:rzr rwrLrxr$) rrrrrrcrhr{r|r}r~rrtr"common_completion_suffix_validatorPs^       (  rcCsdddddd}d}d}d}|jjdddksH|jjdddkrTd }d }|}td |||d S) z This validator will suggest to add a space at the start of the completion if it doesn't already exist. This helps with tokenization. rr>cSs|ddd|d<|S)Nr:cSs|drdnd|S)Nrr!) startswith)rJrrrrErFzLcompletions_space_start_validator..add_space_start..)rKr@rrradd_space_startsz:completions_space_start_validator..add_space_startNrYrrz - The completion should start with a whitespace character (` `). This tends to produce better results due to the tokenization we use. See https://platform.openai.com/docs/guides/fine-tuning/preparing-your-dataset for more detailsz=Add a whitespace character to the beginning of the completionZcompletion_space_startrT)r:r nuniquevaluesr )rrrrrrrr!completions_space_start_validators,rrzRemediation | Noner'csvdddfdd }|dd}|dd}|d|krrtd d d d d d|dSdS)zt This validator will suggest to lowercase the column values, if more than a third of letters are uppercase. rr>cs|j|<|Sr4r)r@r.rr lower_casesz(lower_case_validator..lower_casecSstdd|DS)Ncss"|]}|r|rdVqdSrYN)isalphaisupperr+rrr s9lower_case_validator....sumr@rrrrErFz&lower_case_validator..cSstdd|DS)Ncss"|]}|r|rdVqdSr)rislowerr+rrrrsrrr@rrrrErFr rz - More than a third of your `z%` column/key is uppercase. Uppercase zs tends to perform worse than a mixture of case encountered in normal language. We recommend to lower case the data if that makes sense in your domain. See https://platform.openai.com/docs/guides/fine-tuning/preparing-your-dataset for more detailsz'Lowercase all your data in column/key `r7rTN)rKrr )rr(rZ count_upperZ count_lowerrr.rlower_case_validators  rz'tuple[pd.DataFrame | None, Remediation])fnamer;rc Csd}d}d}d}d}tj|rz6|dsB|dr|drTdnd\}}d|d}d|d }tj||td d }n|d rd }d}t |} | j } t | dkr|d7}tj |tdd }nx|dr@d}d}t |d4} | } tjdd| dD|tdd }W5QRXn|drtj|dtdd }t |dkrXd}d}tj|tdd }nn|drzFtj|dtdd }t |dkrtj|tdd }nd }d}Wn*tk rtj|tdd }YnXn@d!}d"|krH|d#|d$|d"d%d&7}n|d#|d'7}WnDttfk r|d"d%}d(|d)|d*|d+}YnXn d,|d-}td.|||d/}||fS)0z This function will read a file saved in .csv, .json, .txt, .xlsx or .tsv format using pandas. - for .xlsx it will read the first sheet - for .txt it will assume completions and split on newline Nz.csvz.tsv)ZCSV,)ZTSV z= - Based on your file extension, your file is formatted as a z filez Your format `z` will be converted to `JSONL`)sepdtyper!z.xlsxzH - Based on your file extension, your file is formatted as an Excel filez/Your format `XLSX` will be converted to `JSONL`rYz - Your Excel file contains more than one sheet. Please either save as csv or ensure all data is present in the first sheet. WARNING: Reading only the first sheet...)rz.txtz9 - Based on your file extension, you provided a text filez.Your format `TXT` will be converted to `JSONL`rcSsg|] }d|gqS)r!r)r,linerrrr/sz#read_any_format..r\)r0r.jsonlT)linesrz^ - Your JSONL file appears to be in a JSON format. Your file will be converted to JSONL formatz/Your format `JSON` will be converted to `JSONL`z.jsonz^ - Your JSON file appears to be in a JSONL format. Your file will be converted to JSONL formatz]Your file must have one of the following extensions: .CSV, .TSV, .XLSX, .TXT, .JSON or .JSONLrz Your file `z` ends with the extension `.z` which is not supported.z` is missing a file extension.z Your file `z!` does not appear to be in valid z9 format. Please ensure your file is formatted as a valid z file.zFile z does not exist.read_any_format)r rrr)ospathisfiler*endswithpdZread_csvr ZfillnaZ ExcelFileZ sheet_namesr$Z read_excelopenreadZ DataFramesplit read_json ValueError TypeErrorupperr ) rr; remediationrrrrZfile_extension_str separatorZxlsZsheetsfcontentrrrrs      "  rcCs,t|}d}|dkr d|d}td|dS)z This validator will infer the likely fine-tuning format of the data, and display it to the user if it is classification. It will also suggest to use ada and explain train/validation split benefits. NrzK - Based on your data it seems like you're trying to fine-tune a model for z - For classification, we recommend you try one of the faster and cheaper models, such as `ada` - For classification, you can estimate the expected model performance by keeping a held out dataset, which is not used for trainingr"r#)rbr )rrcrrrrformat_inferrer_validators  r)rrrcCsb|jdk r2tjd|jd|jdtd|jdk rJtj|j|jdk r^||}|S)zs This function will apply a necessary remediation to a dataframe, or print an error message if one exists. Nz ERROR in z validator: z Aborting...rY) rr]stderrr_r exitrr^r)rrrrrapply_necessary_remediation(s     rbool) input_text auto_acceptrcCs.tj||r tjddStdkS)NzY Tn)r]r^r_inputr*)rrrrraccept_suggestion6s   rztuple[pd.DataFrame, bool])rrrrcCsjd}d|jd}|jdk rBt||rB|jdk s4t||}d}|jdk rbtjd|jd||fS)zc This function will apply an optional remediation to a dataframe, based on the user input. Fz- [Recommended] z [Y/n]: NTz- [Necessary] r\)rrrAssertionErrorrr]r^r_)rrroptional_appliedrrrrapply_optional_remediation>s    rNonecCsrt|}d}|dkr&t|}|d}n|jdd}|d}ddd d d }||d }tjd |ddS)z? Estimate the time it'll take to fine-tune the dataset g?rg ףp= ?T)rNg|?5^?floatr )timercSsh|dkrt|ddS|dkr4t|dddS|dkrPt|dddSt|dddSdS) N<r z secondsiz minutesiQz hoursz days)round)rrrr format_time]sz.estimate_fine_tuning_time..format_timez:Once your model starts training, it'll approximately take z~ to train a `curie` model, and less for `ada` and `babbage`. Queue will approximately take half an hour per job ahead of you. N)rbr$Z memory_usagerr]r^r_)r ft_formatZ expected_timer"sizer time_stringrrrestimate_fine_tuning_timePs   r)rrrcsf|r ddgndg}d}|dkr*d|dndfdd|D}td d |DsX|S|d 7}qdS) NZ_train_validr!rz ()cs,g|]$}tjdd|dqS)rZ _preparedr)rrsplitext)r,rkrZ index_suffixrrr/rsz!get_outfnames..css|]}tj|VqdSr4)rrr)r,rrrrrssz get_outfnames..rY)rL)rrsuffixesiZcandidate_fnamesrrr get_outfnamesmsrztuple[int, object]cCs.|j}d}|dkr&|jjd}||fS)Nr r)r:rZ value_countsrN)r n_classes pos_classrrrget_classification_hyperparamsxs  r)rrany_remediationsrrc Cst|}t|jdd}t|jdd}d}d}|dkrBt||rBd}d} |dd } |dd } t| d krvd | d nd} d }|s|stj d|d| d| d| d t |nt||rzt ||} |rt| dkrd| d krd| dkst d}t t||tt|d}|j|dd}||j}|ddgj| d ddddd|ddgj| ddddddt|\}}| d7} |dkr| d |d7} n| d!|7} n2t| dkst |ddgj| d ddddd|rd"ndd#d$| }|rd%| ddnd}t| d kr2dn d&| d}tj d'|d(| d d|| d)|| d t |n tj d*dS)+aQ This function will write out a dataframe to a file, if the user would like to proceed, and also offer a fine-tuning command with the newly created file. For classification it will optionally ask the user if they would like to split the data into train/valid files, and modify the suggested command to include the valid set. rkroFzQ- [Recommended] Would you like to split into training and validation set? [Y/n]: rTr!r\rgrz Make sure to include `stop=["z;"]` so that the generated texts ends at the expected place.z@ Your data will be written to a new JSONL file. Proceed [Y/n]: zK You can use your file for fine-tuning: > openai api fine_tunes.create -t ""ue After you’ve fine-tuned a model, remember that your prompt has to end with the indicator string `zX` for the model to start generating completions, rather than continuing with the prompt.r ZtrainvalidrYig?*)rZ random_stater9r:recordsN)rZorientZ force_asciiindentz! --compute_classification_metricsz" --classification_positive_class "z --classification_n_classes rJz to `z` and `z -v "ucAfter you’ve fine-tuned a model, remember that your prompt has to end with the indicator string `z Wrote modified filezd` Feel free to take a look! Now use that file when fine-tuning: > openai api fine_tunes.create -t "z z#Aborting... did not write the file )rbryr9r:rrxr$r]r^r_rrrmaxintsampler`rNZto_jsonrrU)rrrrrZcommon_prompt_suffixrurrZadditional_paramsZ%common_prompt_suffix_new_line_handledZ)common_completion_suffix_new_line_handledZoptional_ending_stringfnamesZMAX_VALID_EXAMPLESZn_trainZdf_trainZdf_validrrZ files_stringZ valid_stringZseparator_reminderrrrwrite_out_files       (        ( rcCs>d}t|jjdkrdSt|jt||kr:dSdS)z> Infer the likely fine-tuning task type from the data rrWrzconditional generation)rr9r r$r:unique)rZCLASSIFICATION_THRESHOLDrrrrbs rbrk)seriesrprcCsld}|dkr$|jt|d dn|jdt|d}|dkrJqhq||jdkr\qhq|jd}q|S)zQ Finds the longest common suffix or prefix of all the values in a series r!rkrYNr)r r$rr)rrpZ common_xfixZ common_xfixesrrrrys4  ryz,Callable[[pd.DataFrame], Remediation | None]r Validatorzlist[Validator])rcCs2tddddtttttddddtttt t gS)NcSs t|dSrlr8r@rrrrErFz get_validators..cSs t|dSrrr@rrrrErFcSs t|dSrlrr@rrrrErFcSs t|dSrrr@rrrrErF) r%rCrPrrVrdrrrrrrrrrget_validatorss rzCallable[..., Any])rrr validatorsrwrite_out_file_funcrc Csg}|dk r|||D](}||}|dk r||t||}qtdd|D}tdd|D} d} |rtjd|D]}t|||\}} | p| } qn tjd| p| } |||| |dS)NcSs$g|]}|jdk s|jdk r|qSr4)rrr,rrrrr/s z$apply_validators..cSsg|]}|jdk r|qSr4)rrrrrr/s Fz? Based on the analysis we will perform the following actions: z No remediations found. )appendrrLr]r^r_r) rrrrrrZoptional_remediationsZ validatorZ&any_optional_or_necessary_remediationsZany_necessary_appliedZany_optional_appliedrZ!any_optional_or_necessary_appliedrrrapply_validatorss2       r)r:)rk), __future__rrr]typingrrrrrtyping_extensionsrZ_extrasr rr rr%r8rCrPrVrdrrrrrrrrrrrrrrrrbryrrrrrrrrsF    $%D'$DY  K