224 lines
49 KiB
Plaintext
224 lines
49 KiB
Plaintext
|
§
|
|||
|
|
4$�hᶠã ó — d Z ddlmZ ddlZddlmZ ddlmZ ddlmZm Z m
|
|||
|
|
Z
|
|||
|
|
m Z m Z m
|
|||
|
|
Z
|
|||
|
|
mZmZ ddlmZmZmZmZmZmZmZmZmZmZmZmZmZmZm
|
|||
|
|
Z
|
|||
|
|
m Z m Z m Z m Z m Z m!Z!m Z m"Z"m#Z#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)m*Z*mZm+Z+m,Z,m-Z-mZm.Z.m/Z/ ddl+Z+ddlT ddl0m1Z1m2Z2 dd l3m4Z4 ddlZddl5Z6dd
|
|||
|
|
l7m8Z8 ddlmZ dd l9m:Z:m;Z< d d
|
|||
|
|
d d
|
|||
|
|
d
|
|||
|
|
dœZ= ej> d d e=¬¦ « d„ ¦ « Z?e1 G d„ de¦ « ¦ « Z@ G d„ de¦ « ZA G d„ deA¦ « ZBdS )z8
|
|||
|
|
2025.8.4
|
|||
|
|
2025.8.5
|
|||
|
|
4.55.1
|
|||
|
|
0.21.0
|
|||
|
|
__UNSLOTH_VERSIONING__
|
|||
|
|
é )ÚTensorN)Ú
|
|||
|
|
functional)ÚAnyÚListÚOptionalÚTupleÚUnionÚDictÚSetÚCallable)&r ÚBaseImageProcessorÚBasePairwiseJudger ÚDatasetÚEvalPredictionÚFÚFeatureExtractionMixinÚGeometricMixtureWrapperÚIterableDatasetÚ NashMDConfigÚ
|
|||
|
|
NashMDTrainerÚOnlineDPOTrainerÚOptimizerNamesr Ú PeftModelÚPreTrainedModelÚPreTrainedTokenizerBaseÚProcessorMixinÚSIMPLE_CHAT_TEMPLATEÚTrainerCallbackr Ú empty_cacheÚgenerate_model_cardÚget_comet_experiment_urlÚ
|
|||
|
|
get_rewardÚis_conversationalÚis_peft_availableÚis_wandb_availableÚjinja2Úmaybe_apply_chat_templateÚnnÚosÚselective_log_softmaxÚtextwrapÚtorchÚtruncate_rightÚunwrap_model_for_generation)Ú*)Ú dataclassÚfield)ÚVersion)Ú nullcontext)ÚDataCollatorForSeq2SeqÚ DataCollatorForLanguageModelingTF)Úepilogue_fusionÚ max_autotuneÚ
|
|||
|
|
shape_paddingz
|
|||
|
|
trace.enabledztriton.cudagraphs)ÚdynamicÚ fullgraphÚoptionsc ó’ — t j | d| j d ¦ « dd¬¦ « }t j | d¦ « dd¬¦ « }g }t ||¦ « D ]‘\ }}| t j ¦ « }t j |d| d¦ « ¬¦ « d¦ « }t j
|
|||
|
|
|d¬¦ « }||z
|
|||
|
|
} | | ¦ « Œ’ t j |¦ « }| | j d | j d f¦ « }|S )Néÿÿÿÿé r )ÚchunksÚdim)r@ Úindex©r@ é )
|
|||
|
|
r, ÚchunkÚreshapeÚshapeÚzipÚtoÚfloat32ÚgatherÚ unsqueezeÚsqueezeÚ logsumexpÚappendÚconcat)
|
|||
|
|
ÚlogitsrA Úchunked_logitsÚ
|
|||
|
|
chunked_indexÚall_per_token_logpsÚ chunk_logitsÚ chunk_indexÚselected_logitsÚlogsumexp_valuesÚper_token_logpss
|
|||
|
|
ú`/workspace/Fine-tuning/DS-LLM-TEMPLATE-FINETUNING/unsloth_compiled_cache/UnslothNashMDTrainer.pyÚ chunked_selective_log_softmaxrZ " s5 € õ ”[ § ¢ °°F´LÀÔ4DÑ!EÔ!EÐPQÐYZÐ[Ñ[Ô[€NÝ”[ § ¢ ¨rÑ!2Ô!2¸QÀaÐHÑHÔH€MØ Ðå%(¨ ¸ Ñ%GÔ%Gð 4ð 4Ñ!ˆ �kØ#—’¥u¤}Ñ5Ô5ˆ Ý œ, |¸2À{×G\ÒG\Ð]_ÑG`ÔG`ÐaÑaÔa×iÒiÐjlÑmÔmˆÝ œ?¨<¸rÐBÑBÔBÐØ)Ð,<Ñ<ˆØ×"Ò" ?Ñ3Ô3Ð3Ð3ØÝ œ,Ð':Ñ;Ô;ÐØ-×5Ò5°v´|ÀA´ÈÌ ÐUVÌÐ6XÑYÔYÐØ Ð ó c ó® ‡ — e Zd ZU dZ edddi¬¦ « Zee ed< edddi¬¦ « Z ee
|
|||
|
|
ed < d/ˆ fd.„ Z ˆ xZ S )0ÚUnslothNashMDConfigaö
|
|||
|
|
|
|||
|
|
Configuration class for the [`NashMDTrainer`].
|
|||
|
|
|
|||
|
|
Subclass of [`OnlineDPOConfig`] we can use all its arguments and add the following:
|
|||
|
|
|
|||
|
|
Parameters:
|
|||
|
|
mixture_coef (`float` or `list[float]`, *optional*, defaults to `0.5`):
|
|||
|
|
Logit mixture coefficient for the model and reference model. If a list of floats is provided then the
|
|||
|
|
mixture coefficient is selected for each new epoch and the last coefficient is used for the rest of the
|
|||
|
|
epochs.
|
|||
|
|
|
|||
|
|
NÚhelpzvLLM SamplingParams)ÚdefaultÚmetadataÚvllm_sampling_paramsr= z8Chunk size to reduce memory usage. -1 is most efficient.Úunsloth_num_chunksFÚnor> é r éú ç-C ëâ6
|
|||
|
|
?ç{®Gáz„?çÍÌÌÌÌÌì?ç+‡ÙÎ÷ï?ç:Œ0âŽyE>ç ð?ç @Úlinearçš™™™™™¹?ÚpassiveÚwarningTÚstepsrC éô éO
|
|||
|
|
ÚO1ÚautoÚ ç Ú
|
|||
|
|
adamw_8bitÚlengthÚ
|
|||
|
|
every_saveÚlasté é@ é ÚsigmoidÚvllmçš™™™™™á?c‘ óÌ •— |dk rt d|› d�¦ « ‚|dk rt d|› d�¦ « ‚|€|#dk r
|
|||
|
|
|$dk rd}d }#|ˆ€!d
|
|||
|
|
d lm}’ t |’¦ « d z d ¦ « }ˆ|…d
|
|||
|
|
k rt d
|
|||
|
|
¦ « ‚|…dk rt d¦ « ‚ t
|
|||
|
|
¦ « j dži d|“d|“d|“d|“d|“d|“d|“d|“d| “d|
|
|||
|
|
“d| “d| “d |
|
|||
|
|
“d |“d |“d |“d |“d!|“d"|“d#|“d$|“d%|“d&|“d'|“d(|“d)|“d*|“d+| “d,| “d-| “d.| “d/| “d0|!“d1|"“d2|#“d3|$“d4|%“d5|&“d6|'“d7|(“d8|)“d9|*“d:|+“d;|,“d<|-“d=|.“d>|/“d?|0“d@|1“dA|2“dB|3“dC|4“dD|5“dE|6“dF|7“dG|8“dH|9“dI|:“dJ|;“dK|<“dL|=“dM|>“dN|?“dO|@“dP|A“dQ|B“dR|C“dS|D“dT|E“dU|F“dV|G“dW|H“dX|I“dY|J“dZ|K“d[|L“d\|M“d]|N“d^|O“d_|P“d`|Q“da|R“db|S“dc|T“dd|U“de|V“df|W“dg|X“dh|Y“di|Z“dj|[“dk|\“dl|]“dm|^“dn|_“do|`“dp|a“dq|b“dr|c“ds|d“dt|e“du|f“dv|g“dw|h“dx|i“dy|j“dz|k“d{|l“d||m“d}|n“d~|o“d|p“d€|q“d�|r“d‚|s“dƒ|t“d„|u“d…|v“d†|w“d‡|x“dˆ|y“d‰|z“dŠ|{“d‹||“dŒ|}“d�|~“dŽ|“d�|€“d�|�“d‘|‚“d’|ƒ“d“|„“d”|…“d•|†“d–|‡“d—|ˆ“d˜|‰“d™|Š“dš|‹“d›|Œ“dœ|�“d�|Ž“|‘¤Ž |�| _ |�| _ d S )ŸNgH¯¼šò×z>z Unsloth: Your learning rate of `zi` is too small and less than 1e-7! Consider increasing it, otherwise gradient updates will be close to 0!rC za` is way too larger > 1! Consider decreasing it to 1e-1, otherwise gradient updates will explode!rq rr Ú unsloth_training_checkpointsrc r )Ú cpu_countrd zUUnsloth: Please set a positive non-zero temperature since your results will be wrong.é
|
|||
|
|
zgUnsloth: Please set a positive non-zero temperature less than 10, since sampling will be quite erratic.Ú
|
|||
|
|
output_dirÚoverwrite_output_dirÚdo_trainÚdo_evalÚ
|
|||
|
|
do_predictÚ
|
|||
|
|
eval_strategyÚprediction_loss_onlyÚper_device_train_batch_sizeÚper_device_eval_batch_sizeÚper_gpu_train_batch_sizeÚper_gpu_eval_batch_sizeÚgradient_accumulation_stepsÚeval_accumulation_stepsÚ
|
|||
|
|
eval_delayÚtorch_empty_cache_stepsÚ
|
|||
|
|
learning_rateÚ weight_decayÚ
|
|||
|
|
adam_beta1Ú
|
|||
|
|
adam_beta2Ú adam_epsilonÚ
|
|||
|
|
max_grad_normÚnum_train_epochsÚ max_stepsÚlr_scheduler_typeÚ warmup_ratioÚ warmup_stepsÚ log_levelÚlog_level_replicaÚlog_on_each_nodeÚ logging_dirÚlogging_strategyÚlogging_first_stepÚ
|
|||
|
|
logging_stepsÚlogging_nan_inf_filterÚ
|
|||
|
|
save_strategyÚ
|
|||
|
|
save_stepsÚsave_total_limitÚsave_safetensorsÚsave_on_each_nodeÚsave_only_modelÚ'restore_callback_states_from_checkpointÚno_cudaÚuse_cpuÚuse_mps_deviceÚseedÚ data_seedÚ
|
|||
|
|
jit_mode_evalÚuse_ipexÚbf16Úfp16Úfp16_opt_levelÚhalf_precision_backendÚbf16_full_evalÚfp16_full_evalÚtf32Ú
|
|||
|
|
local_rankÚ ddp_backendÚ
|
|||
|
|
tpu_num_coresÚtpu_metrics_debugÚdebugÚdataloader_drop_lastÚ
|
|||
|
|
eval_stepsÚdataloader_num_workersÚdataloader_prefetch_factorÚ
|
|||
|
|
past_indexÚrun_nameÚ disable_tqdmÚremove_unused_columnsÚ label_namesÚload_best_model_at_endÚmetric_for_best_modelÚgreater_is_betterÚignore_data_skipÚfsdpÚfsdp_min_num_paramsÚ fsdp_configÚ"fsdp_transformer_layer_cls_to_wrapÚaccelerator_configÚ deepspeedÚlabel_smoothing_factorÚoptimÚ
|
|||
|
|
optim_argsÚ adafactorÚgroup_by_lengthÚlength_column_nameÚ report_toÚddp_find_unused_parametersÚddp_bucket_cap_mbÚddp_broadcast_buffersÚdataloader_pin_memoryÚ dataloader_persistent_workersÚskip_memory_metricsÚuse_legacy_prediction_loopÚ push_to_hubÚresume_from_checkpointÚ hub_model_idÚ hub_strategyÚ hub_tokenÚhub_private_repoÚhub_always_pushÚ hub_revisionÚgradient_checkpointingÚ gradient_checkpointing_kwargsÚinclude_inputs_for_metricsÚeval_do_concat_batchesÚ fp16_backendÚpush_to_hub_model_idÚpush_to_hub_organizationÚpush_to_hub_tokenÚ
|
|||
|
|
mp_parametersÚauto_find_batch_sizeÚfull_determinismÚ torchdynamoÚ ray_scopeÚ ddp_timeoutÚ
|
|||
|
|
torch_compileÚtorch_compile_backendÚtorch_compile_modeÚinclude_tokens_per_secondÚ include_num_input_tokens_seenÚneftune_noise_alphaÚoptim_target_modulesÚbatch_eval_metricsÚ
|
|||
|
|
eval_on_startÚuse_liger_kernelÚliger_kernel_configÚeval_use_gather_objectÚ average_tokens_across_devicesÚreward_model_pathÚjudgeÚmax_new_tokensÚ
|
|||
|
|
max_lengthÚ temperatureÚmissing_eos_penaltyÚ loss_typeÚdataset_num_procÚdisable_dropoutÚuse_vllmÚvllm_model_implÚgpu_memory_utilizationÚds3_gather_for_generationÚmodel_init_kwargs© )
|
|||
|
|
ÚFloatingPointErrorÚ
|
|||
|
|
OverflowErrorÚmultiprocessingr„ ÚminÚ MathErrorÚsuperÚ__init__ra rb )”Úselfr† r‡ rˆ r‰ rŠ r‹ rŒ r� rŽ r� r� r‘ r’ r“ r” r• r– r— r˜ r™ rš r› rœ r� rž rŸ r r¡ r¢ r£ r¤ r¥ r¦ r§ r¨ r© rª r« r¬ r r® r¯ r° r± r² r³ r´ rµ r¶ r· r¸ r¹ rº r» r¼ r½ r¾ r¿ rÀ rÁ r rà rÄ rÅ rÆ rÇ rÈ rÉ rÊ rË rÌ rÍ rÎ rÏ rÐ rÑ rÒ rÓ rÔ rÕ rÖ r× rØ rÙ rÚ rÛ rÜ rÝ rÞ rß rà rá râ rã rä rå ræ rç rè ré rê rë rì rí rî rï rð rñ rò ró rô rõ rö r÷ rø rù rú rû rü rý rþ rÿ r r r r r r r r r r r
|
|||
|
|
r r r
|
|||
|
|
r r r r r r ra rb Úkwargsr„ Ú __class__s” €rY r z UnslothNashMDConfig.__init__J s¿ ø€ ðh ˜4Ò Ð Õ'9ð ;VÐ]jð ;Vð ;Vð ;Vñ (Wô (Wð "WØ ˜1Ò Ð ¥Mð 3FÐUbð 3Fð 3Fð 3Fñ %Gô %Gð GØ Ð -°7Ò":Ð":¸zÈSÒ?PÐ?PØ7ˆJØ ˆMØ Ð #Ø 1Ð 1Ð 1Ð 1Ð 1Ð 1Ý " 9 9¡;¤;¨q¡=°!Ñ 4Ô 4Ð Ø ˜!Ò Ð ÝÐ sÑtÔtÐ tØ
|
|||
|
|
˜BÒ
|
|||
|
|
Ð
|
|||
|
|
Ýð Fñ Gô Gð
|
|||
|
|
Gð �‰ŒÔð N <ð N <ð N <Ø#˜ðN <à#7Ð#7ðN <ð �xðN <ð �gð N <ð
|
|||
|
|
$˜ð N <ð *˜Mð
|
|||
|
|
N <ð $8Ð#7ðN <ð +FÐ*EðN <ð *DÐ)CðN <ð (@Ð'?ðN <ð '>Ð&=ðN <ð +FÐ*EðN <ð '>Ð&=ðN <ð $˜ð N <ð '>Ð&=ð N <ð *˜Mð!N <ð" (˜<ð#N <ð$ $˜ð%N <ð& $˜ð'N <ð( (˜<ð)N <ð* *˜Mð+N <ð, 0Ð /ð-N <ð. "˜ ð/N <ð0 !2Ð 1ð1N <ð2 (˜<ð3N <ð4 (˜<ð5N <ð6 "˜ ð7N <ð8 !2Ð 1ð9N <ð: 0Ð /ð;N <ð< &˜+ð=N <ð> 0Ð /ð?N <ð@ "4Ð!3ðAN <ðB *˜MðCN <ðD &<Ð%;ðEN <ðF *˜MðGN <ðH $˜ðIN <ðJ 0Ð /ðKN <ðL 0Ð /ðMN <ðN !2Ð 1ðON <ðP .˜oðQN <ðR 7^Ð6]ðSN <ðT �gðUN <ðV �gðWN <ðX ,˜^ðYN <ðZ �4ð[N <ð\ "˜ ð]N <ð^ *˜Mð_N <ð` �xðaN <ðb �4ðcN <ðd �4ðeN <ðf ,˜^ðgN <ðh &<Ð%;ðiN <ðj ,˜^ðkN <ðl ,˜^ðmN <ðn �4ðoN <ðp $˜ðqN <ðr &˜+ðsN <ðt *˜MðuN <ðv !2Ð 1ðwN <ðx �EðyN <ðz $8Ð#7ð{N <ð| $˜ð}N <ð~ &<Ð%;ðN <ð@ *DÐ)CðAN <ðB $˜ðCN <ðD �xðEN <ðF (˜<ðGN <ðH %:Ð$9ðIN <ðJ &˜+ðKN <ðL &<Ð%;ðMN <ðN %:Ð$9ðON <ðP !2Ð 1ðQN <ðR 0Ð /ðSN <ðT �4ðUN <ðV #6Ð"5ðWN <ðX &˜+ðYN <ðZ 2TÐ1Sð[N <ð\ "4Ð!3ð]N <ð^ "˜ ð_N <ð` &<Ð%;ðaN <ðb �EðcN <ðd $˜ðeN <ðf "˜ ðgN <ðh .˜oðiN <ðj "4Ð!3ðkN <ðl "˜ ðmN <ðn *DÐ)CðoN <ðp !2Ð 1ðqN <ðr %:Ð$9ðsN <ðt %:Ð$9ðuN <ðv -JÐ,IðwN <ðx #6Ð"5ðyN <ðz *DÐ)Cð{N <ð| &˜+ð}N <ð~ &<Ð%;ðN <ð@ (˜<ðAN <ðB (˜<ðCN <ðD "˜ ðEN <ðF 0Ð /ðGN <ðH .˜oðIN <ðJ (˜<ðKN <ðL &<Ð%;ðMN <ðN -JÐ,IðON <ðP *DÐ)CðQN <ðR &<Ð%;ðSN <ðT (˜<ðUN <ðV $8Ð#7ðWN <ðX (@Ð'?ðYN <ðZ !2Ð 1ð[N <ð\ *˜Mð]N <ð^ $8Ð#7ð_N <ð` 0Ð /ðaN <ðb &˜+ðcN <ðd "˜ ðeN <ðf &˜+ðgN <ðh *˜MðiN <ðj %:Ð$9ðkN <ðl "4Ð!3ðmN <ðn )BÐ(AðoN <ðp -JÐ,IðqN <ðr #6Ð"5ðsN <ðt $8Ð#7ðuN <ðv "4Ð!3ðwN <ðx *˜MðyN <ðz 0Ð /ð{N <ð| #6Ð"5ð}N <ð~ &<Ð%;ðN <ð@ -JÐ,IðAN <ðB !2Ð 1ðCN <ðD �EðEN <ðF ,˜^ðGN <ðH $˜ðIN <ðJ &˜+ðKN <ðL #6Ð"5ðMN <ðN "˜ ðON <ðP 0Ð /ðQN <ðR .˜oðSN <ðT �xðUN <ðV .˜oðWN <ðX &<Ð%;ðYN <ðZ )BÐ(Að[N <ð\ !2Ð 1°Fð]N <ð N <ð N <ð^ %9ˆÔ!Ø"4ˆÔ Ð Ð r[ )�NNFFFrc Fr> r> NNrd rd r re rf rg rh ri rj rk rl r= rm rn r ro rp TNrq FrC Frq rr NTFFFFFFrs rs FFFFrt ru FFNr= NNFrv FNr Nr= NNTNFNNFrv r NNNNrw rx NFFry NNNNTFTFFNNrz NNFNFNFTru NNNrv TFNr{ r| FNNFFNNFFFNFTNNr} r~ rh Nr NTFr€ r� TNNr= )
|
|||
|
|
Ú__name__Ú
|
|||
|
|
__module__Ú __qualname__Ú__doc__r1 ra r r Ú__annotations__rb Úintr Ú
|
|||
|
|
__classcell__©r s @rY r] r] 3 sz ø€ € € € € € ð ð ð +0¨%ØØÐ 1Ð2ð+ñ +ô +И( 3œ-ð ð ñ ð */¨ØØÐ VÐWð*ñ *ô *И #œ ð ð ñ ð Ø #ØØØØ Ø $Ø&'Ø%&Ø#'Ø"&Ø&'Ø"#ØØ"%Ø ØØØØ ØØ ØØ $ØØØ Ø %Ø ØØ"Ø "ØØ!&Ø ØØ Ø Ø !Ø Ø27ØØØ ØØØ ØØØØ Ø!'Ø Ø ØØØØ Ø !ØØ $ØØ!"Ø%)ØØØØ $ØØ!&Ø $Ø Ø ØØ ØØ-1Ø !ØØ!$Ø ØØØ Ø %ØØ%)Ø Ø $Ø $Ø(-Ø "Ø%*ØØ!%ØØ#ØØ Ø ØØ!&Ø(,Ø%*Ø!%Ø Ø #Ø#'Ø ØØ #Ø ØØØØ Ø $Ø !Ø$)Ø(-Ø "Ø #Ø "Ø Ø Ø "Ø!&Ø(,Ø ØØØØØ "Ø Ø Ø ØØ Ø!%Ø$(Ø Ø #Ø ðcr5ð r5ð r5ð r5ð r5ð r5ð r5ð r5ð r5ð r5r[ r] c ó" ‡ — e Zd ZdZddgZ d%deeej f deeej f deeej df d e e
|
|||
|
|
d
|
|||
|
|
e e d e e d e ee
|
|||
|
|
ef d
|
|||
|
|
e ee
|
|||
|
|
eee
|
|||
|
|
f f de eeeeef de e de e egef de ee deej j ej j j f de e ej ej gej f ddf ˆ fd„
|
|||
|
|
Z e d„ ¦ « Z!d„ Z"d„ Z#d„ Z$d„ Z%d„ Z&d „ Z' dd „Z( d&dej d eeeej e)f f d e e* dej fd „Z+ d'd!e e d"e e d#eeee df fd$„Z,ˆ xZ-S )(Ú_UnslothNashMDTrainerrv Útrlznash-mdN©NNÚmodelÚ ref_modelÚ reward_modelr ÚargsÚ
|
|||
|
|
data_collatorÚ
|
|||
|
|
train_datasetÚ eval_datasetÚprocessing_classÚ peft_configÚcompute_metricsÚ callbacksÚ
|
|||
|
|
optimizersÚ preprocess_logits_for_metricsÚreturnc óì •— t ¦ « ||||||||| | |
|
|||
|
|
| | |
|
|||
|
|
|¬¦ « | j j | _ g g g g g g g g g g g g dœ | _ | j �g | j d< g | j d< d S d S )N)r+ r, r- r r. r/ r0 r1 r2 Úreward_processing_classr3 r4 r5 r6 r7 ) úloss/klúobjective/entropyú
|
|||
|
|
loss/scoreúrewards/probabilitiesúrewards/accuraciesúrewards/marginsú logps/chosenúlogps/rejectedúval/model_contain_eos_tokenúval/ref_contain_eos_tokenÚbetaÚ mixture_coefúrewards/chosenúrewards/rejected)r r r. rF Ú
|
|||
|
|
_mixture_coefÚstatsr- )r r+ r, r- r r. r/ r0 r1 r2 r3 r4 r5 r6 r7 r s €rY r z _UnslothNashMDTrainer.__init__„ sÉ ø€ õ& ‰Œ×ÒØØ Ø%ØØØ'Ø'Ø%Ø -Ø$4Ø#Ø +Ø Ø!Ø*Gð ñ
|
|||
|
|
ô
|
|||
|
|
ð
|
|||
|
|
ð$ "œYÔ 3ˆÔð Ø!#Ø Ø%'Ø"$Ø !Ø Ø Ø+-Ø)+ØØ ð
|
|||
|
|
ð
|
|||
|
|
ˆŒ
|
|||
|
|
ð Ô Ð (Ø+-ˆDŒJÐ'Ñ (Ø-/ˆDŒJÐ)Ñ *Ð *Ð *ð )Ð (r[ c óÀ — t | j t ¦ « r>| j j }|t | j ¦ « k r
|
|||
|
|
| j | n | j d S | j S )Nr= )Ú
|
|||
|
|
isinstancerI ÚlistÚstateÚepochÚlen)r rO s rY rF z"_UnslothNashMDTrainer.mixture_coefÀ s\ € å �dÔ($Ñ /Ô /ð &Ø”JÔ$ˆEØ05½¸DÔ<NÑ8OÔ8OÒ0OÐ0O�4Ô% eÔ,Ð,ÐUYÔUgÐhjÔUkÐ kàÔ%Ð %r[ c ó¤ — t || j ¦ « 5 }| |d |d | j ¬¦ « }d d d ¦ « n # 1 swxY w Y | j |¦ « }| j €;t
|
|||
|
|
¦ « r*t |t ¦ « r| ¦ « }n"|}n | j | j ¦ « }t j ¦ « 5 t ||| j | j
|
|||
|
|
| j j ¬¦ « }| |d |d | j ¬¦ « }d d d ¦ « n # 1 swxY w Y ||fS )NÚ input_idsÚattention_mask)rR rS Úgeneration_config)r+ r, rT rF Údevice)r. Ú acceleratorÚgeneraterT Ú unwrap_modelr, r$ rL r Úget_base_modelr, Úno_gradr rF rU ) r r+ ÚpromptsÚ unwrapped_policy_for_gen_ctxÚ model_outputÚpolicy_model_for_gmwÚref_model_for_gmwÚ
|
|||
|
|
mixture_modelÚmixture_outputs rY Ú_generate_completionsz+_UnslothNashMDTrainer._generate_completionsÈ sì € å
|
|||
|
|
(¨°Ô0@Ñ
|
|||
|
|
AÔ
|
|||
|
|
Að ÐEaØ7×@Ò@Ø! +Ô.Ø &Ð'7Ô 8Ø"&Ô"8ð Añ ô ˆLð ð ð ñ ô ð ð ð ð ð ð øøøð ð ð ð ð $Ô /× <Ò <¸UÑ CÔ CÐ ð
|
|||
|
|
Œ>Ð !õ !Ñ"Ô"ð
|
|||
|
|
9¥zÐ2FÍ Ñ'RÔ'Rð
|
|||
|
|
9Ø$8×$GÒ$GÑ$IÔ$IÐ!Ð!ð %9Ð!Ð!ð !%Ô 0× =Ò =¸d¼nÑ MÔ MÐ õ Œ]‰_Œ_ð
|
|||
|
|
ð
|
|||
|
|
Ý 3Ø*Ø+Ø"&Ô"8Ø !Ô .ØÔ'Ô.ð ñ ô ˆMð +× 3Ò 3Ø! +Ô.Ø &Ð'7Ô 8Ø"&Ô"8ð 4ñ ô ˆNð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ñ
|
|||
|
|
ô
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
øøøð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð
|
|||
|
|
ð ˜^Ð+Ð+s$ –*A Á AÁAÃ%AEÅEÅ
|
|||
|
|
Ec ó — |d j d }|d d …|d …f }t || j j | j j ¦ « \ }}t j |d |fd¬¦ « t j |d |fd¬¦ « |d dœ}|d d …|d …f }t || j j | j j ¦ « \ }} t j |d |fd¬¦ « t j |d | fd¬¦ « |d dœ}
|
|||
|
|
||
|
|||
|
|
fS )NrR rC rB rS Úraw©rR rS rd )rF r- r2 Ú eos_token_idÚ pad_token_idr, Úcat) r r] ra r[ Úcontext_lengthÚmodel_completion_idsÚmodel_completion_maskÚ
|
|||
|
|
model_dataÚmixture_completion_idsÚmixture_completion_maskÚ mixture_datas rY Ú_process_completionsz*_UnslothNashMDTrainer._process_completionsø sK € Ø Ô-Ô3°AÔ6ˆð ,¨A¨A¨A¨~¨¨Ð,>Ô ?Ð Ý6DØ $Ô"7Ô"DÀdÔF[ÔFhñ7
|
|||
|
|
ô 7
|
|||
|
|
Ñ3Ð Ð 3õ œ G¨KÔ$8Ð:NÐ#OÐUVÐWÑWÔWÝ #œi¨Ð1AÔ)BÐDYÐ(ZÐ`aÐ bÑ bÔ bؘ5”>ð
|
|||
|
|
ð
|
|||
|
|
ˆ
|
|||
|
|
ð "0°°°°>°?°?Ð0BÔ!CÐ Ý:HØ " DÔ$9Ô$FÈÔH]ÔHjñ;
|
|||
|
|
ô ;
|
|||
|
|
Ñ7Ð Ð 7õ œ G¨KÔ$8Ð:PÐ#QÐWXÐYÑYÔYÝ #œi¨Ð1AÔ)BÐD[Ð(\ÐbcÐ dÑ dÔ dؘ5”>ð
|
|||
|
|
ð
|
|||
|
|
ˆ ð ˜<Ð'Ð'r[ c ó8 — t j ¦ « 5 t | j |d | j j |¦ « \ }}}t | j |d | j j |¦ « \ }}}d d d ¦ « n # 1 swxY w Y | j j �Št j |d | j j k d¬¦ « }t j |d | j j k d¬¦ « }|| xx | j j z cc< || xx | j j z cc< ||fS )NrR r= rB )
|
|||
|
|
r, rZ r" r- r2 rg r. r Úanyrf ) r rl ro ri Ú_Ú model_scoresÚmixture_scoresÚmodel_contain_eosÚmixture_contain_eoss rY Ú_compute_rewardsz&_UnslothNashMDTrainer._compute_rewards so € Ý
|
|||
|
|
Œ]‰_Œ_ð ð Ý!+ØÔ! :¨kÔ#:¸DÔ<QÔ<^Ð`nñ"ô "Ñ ˆAˆ|˜Qõ $.ØÔ! <° Ô#<¸dÔ>SÔ>`Ðbpñ$ô $Ñ ˆAˆ~˜qð ð ð ñ ô ð ð ð ð ð ð øøøð ð ð ð ð Œ9Ô (Ð 4Ý %¤ ¨*°[Ô*AÀTÔEZÔEgÒ*gÐmoÐ pÑ pÔ pÐ Ý"'¤)¨L¸Ô,EÈÔI^ÔIkÒ,kÐqsÐ"tÑ"tÔ"tÐ Ø Ð+Ð+Ð ,Ð ,Ô ,°´ Ô0MÑ MÐ ,Ð ,Ñ ,Ø Ð /Ð/Ð 0Ð 0Ô 0°D´IÔ4QÑ QÐ 0Ð 0Ñ 0à˜^Ð+Ð+s ”AA7Á7A;Á>A;c óº ‡ — |d }| j |d d d …|d …f d¬¦ « }d„ |D ¦ « }| j |d d d …|d …f d¬¦ « }d„ |D ¦ « }t d|d i¦ « rod „ |D ¦ « }t j ¦ « }| t ¦ « Š ˆ fd
|
|||
|
|
„|D ¦ « }ˆ fd „|D ¦ « }d „ |D ¦ « }ˆ fd
|
|||
|
|
„|D ¦ « }| j |t t ||¦ « ¦ « d¬¦ « }t j ||d j ¬¦ « S )Nrd rR T)Úskip_special_tokensc ó6 — g | ]}| ¦ « ‘ŒS r ©Ústrip©Ú.0Ú
|
|||
|
|
completions rY ú
|
|||
|
|
<listcomp>z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>* s$ € Ð!^Ð!^Ð!^¸ *×"2Ò"2Ñ"4Ô"4Ð!^Ð!^Ð!^r[ c ó6 — g | ]}| ¦ « ‘ŒS r r| r~ s rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>/ s$ € Ð#bÐ#bÐ#b¸: J×$4Ò$4Ñ$6Ô$6Ð#bÐ#bÐ#br[ Úpromptr c ó — g | ]}d |dœg‘Œ S ©Ú assistant)ÚroleÚcontentr r~ s rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>1 s0 € ð &ð &ð &ØCM˜+°*Ð=Ð=Ð>ð&ð &ð &r[ c ó< •— g | ]}‰ |¬ ¦ « ‘ŒS ©)Úmessages©Úrender)r ÚmessageÚtemplates €rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>6 s' ø€ ÐPÐPÐP¸W�x—’°�Ñ8Ô8ÐPÐPÐPr[ c ó< •— g | ]}‰ |¬ ¦ « ‘ŒS rŠ rŒ ©r r€ r� s €rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>7 s' ø€ Ð%tÐ%tÐ%tÈz h§o¢o¸z oÑ&JÔ&JÐ%tÐ%tÐ%tr[ c ó — g | ]}d |dœg‘Œ S r… r r~ s rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>9 s0 € ð (ð (ð (ØCM˜+°*Ð=Ð=Ð>ð(ð (ð (r[ c ó< •— g | ]}‰ |¬ ¦ « ‘ŒS rŠ rŒ r‘ s €rY r� z8_UnslothNashMDTrainer._compute_judge.<locals>.<listcomp>< s4 ø€ ð (ð (ð (Ø9C�—’¨�Ñ4Ô4ð(ð (ð (r[ )Ú
|
|||
|
|
return_scores)rU )
|
|||
|
|
r2 Ú batch_decoder# r& Ú EnvironmentÚ from_stringr r rM rG r, ÚtensorrU )
|
|||
|
|
r rl ro ri r[ Úmodel_data_completionsÚmixture_data_completionsÚ environmentÚ probabilityr� s
|
|||
|
|
@rY Ú_compute_judgez$_UnslothNashMDTrainer._compute_judge% så ø€ Ø ˜UÔ#ˆØ!%Ô!6×!CÒ!CØ �{Ô # A A A ~ Ð$6Ô 7ÈTð "Dñ "
|
|||
|
|
ô "
|
|||
|
|
Ð ð "_Ð!^ÐG]Ð!^Ñ!^Ô!^Ð à#'Ô#8×#EÒ#EØ ˜Ô % a a a¨ ¨ ¨ Ð&8Ô 9Ètð $Fñ $
|
|||
|
|
ô $
|
|||
|
|
Ð ð $cÐ#bÐIaÐ#bÑ#bÔ#bÐ Ý ˜h¨°¬
|
|||
|
|
Ð 3Ñ 4Ô 4ð ð&ð &ØQgð&ñ &ô &Ð "õ !Ô,Ñ.Ô.ˆKØ"×.Ò.Õ/CÑDÔDˆHØPÐPÐPÐPÈÐPÑPÔPˆGØ%tÐ%tÐ%tÐ%tÐ]sÐ%tÑ%tÔ%tÐ "ð(ð (ØQið(ñ (ô (Ð $ð(ð (ð (ð (ØG_ð(ñ (ô (Ð $ð ”j×&Ò&Ø Ý •Ð+Ð-EÑFÔFÑ GÔ GØ ð 'ñ
|
|||
|
|
ô
|
|||
|
|
ˆ õ
|
|||
|
|
Œ|˜K°
|
|||
|
|
¸;Ô0GÔ0NÐOÑOÔOÐOr[ c ó® ‡— ˆfd„} |||¦ « }t j ¦ « 5 | j €9| ¦ « 5 |||¦ « }d d d ¦ « n # 1 swxY w Y n || j |¦ « }d d d ¦ « n # 1 swxY w Y |d d d …‰d …f dk }| |d¦ « }| |d¦ « }||fS )Nc óª •— | |d |d ¬¦ « }|j d d …‰dz
|
|||
|
|
d…f }t ||d d d …‰d …f ¦ « }|S )NrR rS )rS rC r= )rP r* )ÚmÚdataÚoutputrP Útoken_logprobsri s €rY Úcompute_logprobs_for_datazJ_UnslothNashMDTrainer._compute_logprobs.<locals>.compute_logprobs_for_dataH sp ø€ Ø�Q�t˜KÔ(¸Ð>NÔ9OÐPÑPÔPˆFØ”] 1 1 1 n°qÑ&8¸2Ð&=Ð#=Ô>ˆFÝ 2°6¸4À Ô;LÈQÈQÈQÐP^ÐP_ÐP_ÐM_Ô;`Ñ aÔ aˆNØ!Ð !r[ rS r rw )r, rZ r, Údisable_adapterÚ masked_fill)r r+ rl ri r¤ Úmodel_logprobs_model_dataÚref_logprobs_model_dataÚmodel_padding_masks ` rY Ú_compute_logprobsz'_UnslothNashMDTrainer._compute_logprobsG s´ ø€ ð "ð "ð "ð "ð "ð %>Ð$=¸eÀZÑ$PÔ$PÐ!õ Œ]‰_Œ_ð `ð `ØŒ~Ð%Ø×*Ò*Ñ,Ô,ð [ð [Ø.GÐ.GÈÈzÑ.ZÔ.ZÐ+ð[ð [ð [ñ [ô [ð [ð [ð [ð [ð [ð [øøøð [ð [ð [ð [øð +DÐ*CÀDÄNÐT^Ñ*_Ô*_Ð'ð `ð `ð `ñ `ô `ð `ð `ð `ð `ð `ð `øøøð `ð `ð `ð `ð (Ð(8Ô 9¸!¸!¸!¸^¸_¸_Ð:LÔ MÐQRÒ RÐØ$=×$IÒ$IÐJ\Ð^aÑ$bÔ$bÐ!Ø"9×"EÒ"EÐFXÐZ]Ñ"^Ô"^Ð à)Ð+BÐCÐCs5 ¦ BÁ
|
|||
|
|
AÁ BÁA Á BÁ"A Á#BÂBÂ Bc ó: — |dz
|
|||
|
|
| d¦ « z }t j ¦ « 5 ||z
|
|||
|
|
}| d¦ « }d d d ¦ « n # 1 swxY w Y ||z d¦ « }| j |z |z
|
|||
|
|
}| ¦ « ||fS )Ng à?rC )Úsumr, rZ rE Úmean) r r§ r¨ rœ ÚscoreÚ log_ratioÚ
|
|||
|
|
kl_div_logÚ kl_div_lossÚlosss rY Ú_compute_lossesz%_UnslothNashMDTrainer._compute_losses` sä € ð ˜sÑ"Ð&?×&CÒ&CÀAÑ&FÔ&FÑFˆõ Œ]‰_Œ_ð *ð *Ø1Ð4KÑKˆIØ"Ÿ š qÑ)Ô)ˆJð *ð *ð *ñ *ô *ð *ð *ð *ð *ð *ð *øøøð *ð *ð *ð *ð !Ð#<Ñ<×AÒAÀ!ÑDÔDˆ ð Œy˜;Ñ&¨Ñ.ˆà�yŠy‰{Œ{˜E :Ð-Ð-s ¯AÁAÁ Ac óf ‡ — ˆ fd„} ‰ j d | |¦ « ¦ « ‰ j d | |¦ « ¦ « | d¦ « } | d¦ « }
|
|||
|
|
‰ j d | | ¦ « ¦ « ‰ j d | |
|
|||
|
|
¦ « ¦ « ‰ j �R‰ j d | | ¦ « ¦ « ‰ j d | |
|
|||
|
|
¦ « ¦ « ‰ j d | |¦ « ¦ « | d¦ « }‰ j d
|
|||
|
|
| |¦ « ¦ « | |
|
|||
|
|
z
|
|||
|
|
}‰ j d | |¦ « ¦ « |d k ¦ « }‰ j d
|
|||
|
|
| |¦ « ¦ « |d d d …|d …f ‰ j j k d¬¦ « }|d d d …|d …f ‰ j j k d¬¦ « }‰ j d | | ¦ « ¦ « ¦ « ‰ j d | | ¦ « ¦ « ¦ « ‰ j d ‰ j ¦ « ‰ j d ‰ j ¦ « d S )Nc ó€ •— ‰j | ¦ « ¦ « ¦ « S ©N)rV Úgather_for_metricsr Úitem)r˜ r s €rY Ú gather_meanz:_UnslothNashMDTrainer._log_statistics.<locals>.gather_mean‚ s2 ø€ ØÔ#×6Ò6°vÑ>Ô>×CÒCÑEÔE×JÒJÑLÔLÐ Lr[ r= r; rC rA rB rG rH r> r< r@ r r? rR rB rC rD rE rF )
|
|||
|
|
rJ rN r¬ r- Úfloatr2 rf rr rE rF )r rl ro r§ r¨ rœ r® Úkl_divri rt ru r¹ Ú model_logprobs_model_data_sumÚref_logprobs_model_data_sumÚentropy_model_dataÚmarginÚaccuracyÚ model_eosÚ mixture_eoss ` rY Ú_log_statisticsz%_UnslothNashMDTrainer._log_statisticst s- ø€ ð Mð Mð Mð Mð Mð
|
|||
|
|
Œ
|
|||
|
|
�<Ô ×'Ò'¨ ¨ °EÑ(:Ô(:Ñ;Ô;Ð;à Œ
|
|||
|
|
�9Ô ×$Ò$ [ [°Ñ%8Ô%8Ñ9Ô9Ð9ð )B×(EÒ(EÀaÑ(HÔ(HÐ%Ø&=×&AÒ&AÀ!Ñ&DÔ&DÐ#à Œ
|
|||
|
|
�>Ô"×)Ò)¨+¨+Ð6SÑ*TÔ*TÑUÔUÐUØ Œ
|
|||
|
|
Ð#Ô$×+Ò+¨K¨KÐ8SÑ,TÔ,TÑUÔUÐUð Ô Ð (Ø ŒJÐ'Ô (× /Ò /° ° ¸LÑ0IÔ0IÑ JÔ JÐ JØ ŒJÐ)Ô *× 1Ò 1°+°+¸nÑ2MÔ2MÑ NÔ NÐ Nð
|
|||
|
|
Œ
|
|||
|
|
Ð*Ô+×2Ò2°;°;¸{Ñ3KÔ3KÑLÔLÐLð 8× ;Ò ;¸AÑ >Ô >Ð >ÐØ Œ
|
|||
|
|
Ð&Ô'×.Ò.¨{¨{Ð;MÑ/NÔ/NÑOÔOÐOð /Ð1LÑLˆØ Œ
|
|||
|
|
Ð$Ô%×,Ò,¨[¨[¸Ñ-@Ô-@ÑAÔAÐAð ˜Q’J×%Ò%Ñ'Ô'ˆØ Œ
|
|||
|
|
Ð'Ô(×/Ò/° ° ¸HÑ0EÔ0EÑFÔFÐFð Ô,¨Q¨Q¨Q°°°Ð-?Ô@ÀDÔDYÔDfÒf×kÒkÐpqÐkÑrÔrˆ Ø# KÔ0°°°°N°O°OÐ1CÔDÈÔH]ÔHjÒj×oÒoÐtuÐoÑvÔvˆ Ø Œ
|
|||
|
|
Ð0Ô1×8Ò8¸¸ÀYÇ_Â_ÑEVÔEVÑ9WÔ9WÑXÔXÐXØ Œ
|
|||
|
|
Ð.Ô/×6Ò6°{°{À;×CTÒCTÑCVÔCVÑ7WÔ7WÑXÔXÐXð
|
|||
|
|
Œ
|
|||
|
|
�6Ô×!Ò! $¤)Ñ,Ô,Ð,Ø Œ
|
|||
|
|
�>Ô"×)Ò)¨$Ô*;Ñ<Ô<Ð<Ð<Ð<r[ ÚinputsÚnum_items_in_batchc ó$ ‡ ‡— | ¦ « t t t ‰ ¦ « ¦ « ¦ « ¦ « }‰d }ˆfd„t |¦ « D ¦ « Šˆ fd„‰D ¦ « Šˆ fd„‰D ¦ « Š‰ ‰¦ « Š‰ ‰¦ « Š‰d j d }‰d ‰d |dœ}‹‰ ||¦ « \ }}‰
|
|||
|
|
|||¦ « \ } }
|
|||
|
|
‰ j �2‰ | |
|
|||
|
|
|¦ « \ } } t j | | z
|
|||
|
|
¦ « }
|
|||
|
|
n d \ } } ‰ | |
|
|||
|
|
|¦ « }
|
|||
|
|
‰ || |¦ « \ }}‰ |||
|
|||
|
|
¦ « \ }}}‰ | |
|
|||
|
|
| ¦ « ||
|
|||
|
|
| ¦ « | ¦ « || | ¦
|
|||
|
|
«
|
|||
|
|
‰ j j �+‰ j j ‰ j j z d
|
|||
|
|
k rt1 ¦ « i }‰ j j t4 j t4 j fv r‰ ¦ « |d < ‰ j j dk r| ¦ « }‰ j rMtB " |‰ j# ¦ « 5 }| $ ¦ « d d d ¦ « n # 1 swxY w Y n ‰ j% j$ |fi |¤Ž | ¦ « ‰ j j& z S ) Nrƒ c óR •‡— g | ]"Šˆfd „‰ ¦ « D ¦ « ‘Œ#S )c ó( •— i | ]\ }}||‰ “ŒS r r )r ÚkÚvÚis €rY ú
|
|||
|
|
<dictcomp>zB_UnslothNashMDTrainer.training_step.<locals>.<listcomp>.<dictcomp>· s# ø€ Ð6Ð6Ð6™t˜q !�1�a˜”dÐ6Ð6Ð6r[ )Úitems)r rË rÄ s @€rY r� z7_UnslothNashMDTrainer.training_step.<locals>.<listcomp>· s7 øø€ ÐRÐRÐR¸1Ð6Ð6Ð6Ð6 v§|¢|¡~¤~Ð6Ñ6Ô6ÐRÐRÐRr[ c ó: •— g | ]}t |‰j ¦ « ‘ŒS r )r' r2 ©r Úxr s €rY r� z7_UnslothNashMDTrainer.training_step.<locals>.<listcomp>¸ s' ø€ ÐVÐVÐVÈ!Õ+¨A¨tÔ/DÑEÔEÐVÐVÐVr[ c óf •— g | ]-}‰ |‰j j j ‰j ¦ « ‘Œ.S r )Ú tokenize_rowr+ ÚconfigÚis_encoder_decoderr2 rÏ s €rY r� z7_UnslothNashMDTrainer.training_step.<locals>.<listcomp>¹ s7 ø€ ÐtÐtÐtÐhi�$×#Ò# A t¤zÔ'8Ô'KÈTÔMbÑcÔcÐtÐtÐtr[ Úprompt_input_idsrC Úprompt_attention_maskre r* r r• )'ÚtrainrP ÚnextÚiterÚvaluesÚranger/ Ú_prepare_inputsrF rb rp r- rx r r r� rª r³ rà Údetachr. r” rN Ú global_stepr rÖ r ÚLOMOÚADALOMOÚ_get_learning_rateÚn_gpur Úuse_apexÚampÚ
|
|||
|
|
scale_lossÚ optimizerÚbackwardrV r‘ )r r+ rÄ rÅ Ú
|
|||
|
|
batch_sizer[ ri r] ra rl ro rt ru rœ r§ r¨ r² r® r» r Ú scaled_losss ` ` rY Ú
|
|||
|
|
training_stepz#_UnslothNashMDTrainer.training_step¯ sl øø€ ð � Š ‰
|
|||
|
|
Œ
|
|||
|
|
ˆ
|
|||
|
|
õ ��d 6§=¢=¡?¤?Ñ 3Ô 3Ñ4Ô4Ñ5Ô5ˆ
|
|||
|
|
ؘÔ"ˆØRÐRÐRÐRÅÀjÑ@QÔ@QÐRÑRÔRˆØVÐVÐVÐVÈvÐVÑVÔVˆØtÐtÐtÐtÐmsÐtÑtÔtˆØ×#Ò# FÑ+Ô+ˆð ×%Ò% fÑ-Ô-ˆØ Ð 2Ô3Ô9¸!Ô<ˆà Ð 2Ô3Ø $Ð%<Ô =Øð
|
|||
|
|
ð
|
|||
|
|
ˆð
|
|||
|
|
|
|||
|
|
ð (,×'AÒ'AÀ%ÈÑ'QÔ'QÑ$ˆ �nð $(×#<Ò#<¸\È>Ð[bÑ#cÔ#cÑ ˆ
|
|||
|
|
�Lð Ô Ð (Ø+/×+@Ò+@ÀÈ\Ð[iÑ+jÔ+jÑ (ˆL˜.åœ) L°>Ñ$AÑBÔBˆKˆKà+5Ñ (ˆL˜.Ø ×-Ò-¨j¸,ÈÑWÔWˆKð >B×=SÒ=SÐTYÐ[eÐguÑ=vÔ=vÑ:Ð!Ð#:ð #× 2Ò 2Ð3LÐNeÐgrÑ sÔ sшˆe�Vð
|
|||
|
|
× Ò Ø Ø Ø %× ,Ò ,Ñ .Ô .Ø #Ø Ø �LŠL‰NŒNØ �MŠM‰OŒOØ Ø Ø ñ
|