tokenizer-lm-ablations / ablation_grid.csv
cmeister's picture
Add files using upload-large-folder tool
59eb3e1 verified
Raw
History Blame Contribute Delete
86.5 kB
run,works,training_regime,tokenizer_slug,n_layer,n_embd,n_head,n_kv_head,vocab_size,ve_dim,window_pattern,logit_softcap,sequence_len,total_parameters,step,token_budget,seed,weight_dtype,val_bpb,blimp_acc,code_bpb_mean,flores_mean_bpb,flores_n_languages,mbpp_pass1,humaneval_pass1,gsm8k_exact_match_strict,tok_algorithm,tok_pretokenizer,tok_normalizer,tok_training_data,tok_vocab_total,tok_special_flags,tok_loadable_with,tok_redistributable
full-128k-allmulti-equal-bpe,paper_a;tokeval,std-1B,allmulti-equal-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7176754255270483,0.8189402985074629,0.5210339837815584,2.651776787012788,215,0.056,0.06097560975609756,0.045489006823351025,bpe,REGEX_GPT4O,,allmulti_equal,128260,,transformers,True
full-128k-apertus,tokeval;tokeval_cross_scale,std-1B,swiss-ai/Apertus-70B-2509,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7196643583908005,0.8191044776119404,0.5261638466492652,2.7683366310551216,214,0.086,0.054878048780487805,0.04094010614101592,,,,,,,transformers,False
full-128k-apertus-mathcode-scratch,blog,mathcode-20B-from-scratch,apertus,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3368332422686206,,0.37824387350146926,3.4253459955956593,214,0.0,0.1524390243902439,0.18423047763457165,,,,,,,transformers,False
full-128k-apertus-seed42,paper_a;tokeval_cross_scale,std-1B,swiss-ai/Apertus-70B-2509,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7211964996436109,0.8171194029850744,0.5290844455840652,2.7557483320952856,215,,,,,,,,,,transformers,False
full-128k-bpe-nfc-clean-balanced,paper_a;tokeval,std-1B,bpe-nfc-clean-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.715856631626823,0.8210597014925377,0.5226398359086579,2.6477655259752817,215,0.092,0.08536585365853659,0.03790750568612585,BPE,,NFC,,128004,,transformers,True
full-128k-bpe-nfc-clean-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,bpe-nfc-clean-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33796765404762164,,0.3814762402622109,3.365416060327055,214,0.218,0.15853658536585366,0.177407126611069,BPE,,NFC,,128004,,transformers,True
full-128k-bpe-nfc-clean-balanced-v128k,tokeval,std-1B,bpe-nfc-clean-balanced-v128k,24,1536,12,12,128000,128,SSSL,15.0,2048,1271662346,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.716152463637094,0.8208208955223881,0.5244360004512393,2.644795981112231,215,0.086,0.09146341463414634,0.032600454890068235,bpe,REGEX_CLEAN,NFC,balanced,128004,,transformers,True
full-128k-bpe-nfc-clean-balanced-v160k,tokeval,std-1B,bpe-nfc-clean-balanced-v160k,24,1536,12,12,160000,128,SSSL,15.0,2048,1419118346,8667,9088008192,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7142973980056193,0.8268955223880593,0.5210469295133383,2.6529233096337332,215,0.064,0.08536585365853659,0.039423805913570885,bpe,REGEX_CLEAN,NFC,balanced,160004,,transformers,True
full-128k-bpe-nfc-clean-balanced-v200k,tokeval,std-1B,bpe-nfc-clean-balanced-v200k,24,1536,12,12,200000,128,SSSL,15.0,2048,1603438346,8552,8967421952,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7128314584843901,0.815268656716418,0.5192765421244038,2.669924969239849,215,0.092,0.09146341463414634,0.03866565579984837,bpe,REGEX_CLEAN,NFC,balanced,200004,,transformers,True
full-128k-bpe-nfc-clean-balanced-v64k,tokeval,std-1B,bpe-nfc-clean-balanced-v64k,24,1536,12,12,64000,128,SSSL,15.0,2048,976750346,9299,9750708224,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7193331865186551,0.8131194029850745,0.5268749863541311,2.6168464994006033,215,0.086,0.07317073170731707,0.05686125852918878,bpe,REGEX_CLEAN,NFC,balanced,64004,,transformers,True
full-128k-bpe-nfc-clean-balanced-v96k,tokeval,std-1B,bpe-nfc-clean-balanced-v96k,24,1536,12,12,96000,128,SSSL,15.0,2048,1124206346,8986,9422503936,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7186254102526758,0.8216865671641791,0.5264406583223474,2.644620729776012,215,0.074,0.09146341463414634,0.03639120545868082,bpe,REGEX_CLEAN,NFC,balanced,96004,,transformers,True
full-128k-bpe-nfc-clean-fineweb2full,paper_a,std-1B,bpe-nfc-clean-fineweb2full,24,1536,12,12,128000,128,SSSL,15.0,2048,1271662346,8796,9223274496,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7249761642249047,0.8208358208955225,0.5276609173558784,2.998798131703424,215,0.068,0.07926829268292683,0.02880970432145565,BPE,,NFC,,128000,,transformers,True
full-128k-bpe-nfc-clean-nomark-balanced,paper_a,std-1B,bpe-nfc-clean-nomark-balanced,24,1536,12,12,128000,128,SSSL,15.0,2048,1271662346,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7163375459110471,0.8215223880597016,0.5240401515508685,2.6437112475716065,215,0.09,0.07926829268292683,0.04245640636846096,bpe,REGEX_CLEAN_NOMARK,NFC,balanced,128004,,transformers,True
full-128k-bpe-nfc-gpt4-fineweb2full,paper_a,std-1B,bpe-nfc-gpt4-fineweb2full,24,1536,12,12,128000,128,SSSL,15.0,2048,1271662346,8796,9223274496,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7240783473399763,0.823686567164179,0.5254870925448322,2.9618560118364603,215,0.08,0.07317073170731707,0.0356330553449583,BPE,,NFC,,128000,,transformers,True
full-128k-bpe-nfc-plus2-balanced,paper_a;tokeval,std-1B,bpe-nfc-plus2-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7162315897661191,0.8137164179104476,0.5248969014124836,2.638139141518543,215,0.11,0.0975609756097561,0.029567854435178165,bpe,REGEX_CLEAN_PLUS2,NFC,balanced,128260,,transformers,True
full-128k-bpe-nfc-plus2-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,bpe-nfc-plus2-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3384368951749275,,0.3785201421477085,3.372071226576324,214,0.114,0.1524390243902439,0.21152388172858225,bpe,REGEX_CLEAN_PLUS2,NFC,balanced,128260,,transformers,True
full-128k-bpe-nfc-plus2-balanced-seed43-mathcode-scratch,blog,mathcode-20B-from-scratch,bpe-nfc-plus2-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,43,"as trained (float32 matrices, bfloat16 embeddings)",0.3387550570042015,,,3.3593633851683737,215,0.212,0.17682926829268292,0.20242608036391205,bpe,REGEX_CLEAN_PLUS2,NFC,balanced,128260,,transformers,True
full-128k-bpe-nfc-plus2-balanced-seed44-mathcode-scratch,blog,mathcode-20B-from-scratch,bpe-nfc-plus2-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,44,"as trained (float32 matrices, bfloat16 embeddings)",0.3387273960099328,,,3.374212385169335,215,0.228,0.15853658536585366,0.19863532979529946,bpe,REGEX_CLEAN_PLUS2,NFC,balanced,128260,,transformers,True
full-128k-bpe-nfc-plus3-balanced,paper_a;tokeval,std-1B,bpe-nfc-plus3-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7149187453363852,0.8184626865671641,0.523232159118093,2.6352858885574904,215,0.066,0.08536585365853659,0.039423805913570885,bpe,REGEX_CLEAN_PLUS3,NFC,balanced,128260,,transformers,True
full-128k-bpe-nfc-plus3-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,bpe-nfc-plus3-balanced,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33869831349958357,,0.3834472849799035,3.3760685904096084,214,0.22,0.17073170731707318,0.19711902956785443,bpe,REGEX_CLEAN_PLUS3,NFC,balanced,128260,,transformers,True
full-128k-claude-allmulti-equal-bpe,paper_a;tokeval,std-1B,claude-allmulti-equal-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7199554950159701,0.8167611940298506,0.5231038867012525,2.653515925549718,215,,,,bpe,REGEX_CLAUDE,,allmulti_equal,128260,,transformers,True
full-128k-claude-balanced-bpe,tokeval;tokeval_cross_scale,std-1B,claude-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7139588270432892,0.8212238805970153,0.5184472344138235,2.642247695092226,214,0.076,0.07317073170731707,0.03184230477634572,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,claude-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33648727932470396,,0.3861471966147764,3.3743421228681676,214,0.25,0.17073170731707318,0.22744503411675512,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,claude-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7155515391969438,0.8163134328358209,0.5202755220264306,2.641672000878933,215,0.082,0.07926829268292683,0.03866565579984837,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-bpe-seed43-mathcode-scratch,blog,mathcode-20B-from-scratch,claude-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,43,"as trained (float32 matrices, bfloat16 embeddings)",0.337272087796867,,,3.3792211859742856,215,0.234,0.16463414634146342,0.23199393479909022,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-bpe-seed44-mathcode-scratch,blog,mathcode-20B-from-scratch,claude-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,44,"as trained (float32 matrices, bfloat16 embeddings)",0.33757502590578914,,,3.3411153775837144,215,0.24,0.17073170731707318,0.21455648218347234,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-nfc-bpe,tokeval;tokeval_cross_scale,std-1B,claude-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7135749335882782,0.8230895522388058,0.517861350253653,2.637053701281863,214,0.06,0.08536585365853659,0.03411675511751327,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
full-128k-claude-balanced-nfc-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,claude-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33714380335290617,,0.38451702385863634,3.3666954557164184,215,0.224,0.15853658536585366,0.2304776345716452,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
full-128k-claude-balanced-nfc-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,claude-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7152138612405923,0.8188955223880598,0.520628642499762,2.6415782571300936,215,0.098,0.08536585365853659,0.043214556482183475,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
full-128k-claude-balanced-unigram,tokeval;tokeval_cross_scale,std-1B,claude-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7320967780165727,0.8351641791044776,0.5539673906500296,2.68308108385546,214,0.052,0.07317073170731707,0.027293404094010616,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-unigram-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,claude-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3545650839336789,,0.4032088690906815,3.369774047069325,215,0.222,0.14634146341463414,0.1645185746777862,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-balanced-unigram-seed42,paper_a;tokeval_cross_scale,std-1B,claude-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.735912221128107,0.840686567164179,0.5577539471842974,2.6830926833304374,215,0.058,0.07317073170731707,0.027293404094010616,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
full-128k-claude-english-bpe,tokeval;tokeval_cross_scale,std-1B,claude-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7312714204866293,0.8178805970149255,0.5454915866651372,2.6425029933386486,214,0.068,0.07926829268292683,0.032600454890068235,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
full-128k-claude-english-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,claude-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34048861753468623,,0.36517357096573205,3.1992614684262377,215,0.234,0.16463414634146342,0.221379833206975,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
full-128k-claude-english-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,claude-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7329930803024799,0.8168358208955225,0.5426109795117547,2.6410893400233615,215,0.072,0.06707317073170732,0.032600454890068235,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
full-128k-clauderev-balanced-bpe,paper_a;tokeval,std-1B,clauderev-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7095599606329199,0.8197910447761193,0.520372828134578,2.6486381006261834,215,0.06,0.07317073170731707,0.047763457164518575,bpe,REGEX_CLAUDE_REV,Sequence,balanced,128260,,transformers,True
full-128k-clauderev-balanced-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,clauderev-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33694964163473035,0.743537313432836,0.3561613667750621,3.375535076836162,215,0.238,0.17073170731707318,0.20773313115996966,bpe,REGEX_CLAUDE_REV,Sequence,balanced,128260,,transformers,True
full-128k-gpt2-balanced-bpe,paper_a;tokeval,std-1B,gpt2-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.712764934305536,0.8157761194029848,0.5154497166409894,2.640532228202252,214,0.03,0.07926829268292683,0.027293404094010616,bpe,ByteLevel,,balanced,128260,,transformers,True
full-128k-gpt4o-allmulti-equal-nfc-bpe,paper_a;tokeval,std-1B,gpt4o-allmulti-equal-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7164581029723067,0.8234776119402984,0.5197707064555515,2.6459986350982936,215,,,,bpe,REGEX_GPT4O,NFC,allmulti_equal,128260,,transformers,True
full-128k-gpt4o-allmulti-equal-norepeat-bpe,paper_a;tokeval,std-1B,gpt4o-allmulti-equal-norepeat-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7164844120425224,0.8191791044776121,0.519729913930031,2.647938997818136,215,,,,bpe,REGEX_GPT4O,,allmulti_equal_norepeat,128260,,transformers,True
full-128k-gpt4o-balanced-bpe,tokeval;tokeval_cross_scale,std-1B,gpt4o-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7113825837842997,0.816850746268657,0.5183171305128864,2.6387506982410183,214,0.068,0.09146341463414634,0.045489006823351025,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,gpt4o-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7131214303656023,0.8196268656716418,0.5205519366009381,2.6409485662618386,215,0.066,0.06707317073170732,0.04624715693707354,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-bpe-seed43-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,43,"as trained (float32 matrices, bfloat16 embeddings)",0.3362506080752585,,,3.3809551460810874,215,0.072,0.15853658536585366,0.24184988627748294,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-bpe-seed44-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,44,"as trained (float32 matrices, bfloat16 embeddings)",0.33549283645238004,,,3.3683389882182975,215,0.018,0.16463414634146342,0.23199393479909022,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-maxfiles50-bpe,paper_a,std-1B,gpt4o-balanced-maxfiles50-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7115753975968808,0.8217014925373136,0.5175390204895408,2.6288706791146534,215,,,,bpe,REGEX_GPT4O,,balanced_maxfiles50,128260,,transformers,True
full-128k-gpt4o-balanced-nfc-bpe,tokeval;tokeval_cross_scale,std-1B,gpt4o-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7114993123173433,0.8128955223880601,0.5185776938972276,2.637031235082467,214,0.068,0.07926829268292683,0.04852160727824109,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-nfc-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,gpt4o-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3360314499008519,,0.3786237132757522,3.4023914971799996,214,0.12,0.1524390243902439,0.20773313115996966,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-nfc-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,gpt4o-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7133388823384232,0.8202089552238806,0.520749213831268,2.6376940912300553,215,0.05,0.07926829268292683,0.04700530705079606,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-unigram,tokeval;tokeval_cross_scale,std-1B,gpt4o-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7314636077017965,0.8325522388059705,0.5537964877519875,2.6859967266679656,214,0.066,0.06707317073170732,0.024260803639120546,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-unigram-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,gpt4o-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3533980943471727,,0.3993466898373258,3.4055634092286033,215,0.0,0.13414634146341464,0.1865049279757392,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-unigram-seed42,paper_a;tokeval_cross_scale,std-1B,gpt4o-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7337511969511488,0.8338208955223879,0.5569454278048204,2.6873471717751567,215,0.04,0.07317073170731707,0.032600454890068235,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-gpt4o-balanced-unigram-tuned,paper_a;tokeval,std-1B,gpt4o-balanced-unigram-tuned,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7332765694658377,0.838014925373134,0.5564997843345418,2.686368297404418,215,0.042,0.054878048780487805,0.028051554207733132,unigram,REGEX_GPT4O,,balanced,128260,"{""shrinking_factor"": 0.7, ""n_sub_iterations"": 3, ""max_piece_length"": 64, ""initial_alphabet"": [""\u00d9"", ""e"", ""\u00f3"", ""\u0112"", ""\u0116"", ""\u00bf"", ""\u00ed"", ""\u011f"", ""\u0123"", "">"", ""8"", ""u"", ""\u00ec"", ""s"", ""\u00e7"", ""\u013e"", ""\u0132"", ""\u010e"", ""\u0119"", ""F"", ""G"", ""\u00a8"", ""^"", ""j"", ""-"", ""\u0138"", ""\u0136"", ""\u012a"", ""#"", ""\u00f5"", ""r"", ""\u0137"", ""k"", ""\u0104"", ""m"", ""\u0143"", ""X"", ""3"", ""\u0101"", ""\u00a9"", ""\u00b7"", ""\u00c7"", ""b"", "")"", ""'"", ""\u00ee"", ""\u012c"", ""\u00a3"", ""\u0117"", ""\u0135"", ""4"", ""\u00d1"", ""\u00ca"", ""g"", ""\u00f2"", ""W"", ""\u0111"", ""\u00bc"", ""h"", ""\u00fd"", ""\u00b2"", ""\u00bd"", ""\u00dc"", ""\u0133"", ""\u0121"", ""\u00e8"", ""\u011a"", ""\u00f0"", ""\u00e1"", ""o"", ""l"", ""y"", ""\u0113"", ""\u00c4"", ""6"", ""\u0108"", ""!"", ""\u0126"", ""i"", ""\u00f4"", ""\u00d0"", ""N"", ""\u0120"", ""\u00d7"", ""\u013c"", ""P"", ""\u012d"", ""\u00ae"", ""\u00c6"", ""/"", ""\u00cb"", ""\u011c"", ""A"", ""*"", ""\u00f6"", ""\u00f7"", ""\u0106"", ""$"", ""\u00e5"", ""p"", ""\u00ce"", ""\u0130"", ""\u00ba"", ""\u00fb"", ""\u0105"", ""\u00f9"", ""{"", ""O"", ""\u00d2"", ""Y"", ""Z"", ""Q"", ""\u013d"", ""\u00c2"", ""\u00e0"", ""H"", ""\u010a"", ""z"", ""\u00b3"", ""\u00be"", ""\u00d6"", ""\u00a7"", ""\u0134"", ""\u00af"", ""\u00c1"", ""\u00ef"", ""\u011b"", ""\u0127"", ""\u00b9"", ""\u013f"", ""\u00df"", ""\u010d"", ""D"", ""2"", ""\u00ea"", "":"", ""\u010f"", ""K"", ""\u00ff"", ""a"", ""\\"", ""7"", ""\u0125"", ""\u00f8"", ""\"""", ""\u00a1"", ""\u00b5"", ""\u0110"", ""\u010c"", ""\u00da"", "";"", ""L"", ""\u00fe"", ""c"", ""1"", ""+"", ""\u00ab"", ""\u012b"", ""E"", ""\u0131"", ""\u011e"", ""9"", ""\u0142"", ""<"", ""\u0118"", ""\u00a6"", ""f"", ""\u00c5"", ""M"", ""\u00aa"", ""["", ""\u00cd"", ""\u00bb"", ""\u013a"", ""\u0115"", ""`"", ""("", ""0"", ""v"", "","", ""]"", ""I"", ""."", ""\u0129"", ""\u0141"", ""B"", ""\u00ac"", ""\u00a4"", ""}"", ""T"", ""\u00fa"", ""\u00dd"", ""\u00cc"", ""x"", ""\u00b1"", ""\u00c3"", ""\u00d3"", ""&"", ""t"", ""\u00e2"", ""q"", ""d"", ""\u0103"", ""\u00b4"", ""~"", ""\u00eb"", ""\u00b8"", ""%"", ""="", ""U"", ""\u00b0"", ""\u00d5"", ""\u011d"", ""\u00e4"", ""|"", ""\u010b"", ""V"", ""\u00db"", ""\u0109"", ""\u00e3"", ""_"", ""\u0139"", ""\u012f"", ""@"", ""\u0102"", ""\u012e"", ""\u00fc"", ""\u00c8"", ""w"", ""n"", ""R"", ""\u00c0"", ""\u0122"", ""\u00d4"", ""\u00d8"", ""\u00a5"", ""\u00f1"", ""\u00e6"", ""\u00e9"", ""\u00cf"", ""C"", ""\u00c9"", ""\u0140"", ""?"", ""\u0100"", ""\u0114"", ""\u00b6"", ""J"", ""\u0128"", ""5"", ""\u0124"", ""\u00a2"", ""S"", ""\u00de"", ""\u013b"", ""\u0107""]}",transformers,True
full-128k-gpt4o-code-bpe,tokeval;tokeval_cross_scale,std-1B,gpt4o-code-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7238738916126876,0.821,0.5431368524901159,2.625442837285647,214,0.054,0.07317073170731707,0.039423805913570885,bpe,REGEX_GPT4O,,code,128260,,transformers,True
full-128k-gpt4o-code-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,gpt4o-code-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33264187797540906,,0.3804701231471243,3.152441006253198,214,0.016,0.15853658536585366,0.21607278241091737,bpe,REGEX_GPT4O,,code,128260,,transformers,True
full-128k-gpt4o-code-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,gpt4o-code-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7252439607412663,0.8190149253731343,0.5446616749526122,2.625681797157079,215,0.08,0.09146341463414634,0.0310841546626232,bpe,REGEX_GPT4O,,code,128260,,transformers,True
full-128k-gpt4o-codegrad00-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-codegrad00-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.339566757938572,,,3.357431043588078,215,0.232,0.16463414634146342,0.20545868081880211,bpe,REGEX_GPT4O,,codegrad00,128260,,transformers,True
full-128k-gpt4o-codegrad05-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-codegrad05-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3365866322863168,,,3.3960370708190832,215,0.05,0.1402439024390244,0.21607278241091737,bpe,REGEX_GPT4O,,codegrad05,128260,,transformers,True
full-128k-gpt4o-codegrad30-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-codegrad30-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33496232135285,,,3.3437376646164667,215,0.178,0.15853658536585366,0.20545868081880211,bpe,REGEX_GPT4O,,codegrad30,128260,,transformers,True
full-128k-gpt4o-codegrad50-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-codegrad50-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33476615801535653,,,3.3840800213850546,215,0.088,0.18902439024390244,0.21607278241091737,bpe,REGEX_GPT4O,,codegrad50,128260,,transformers,True
full-128k-gpt4o-english-bpe,tokeval;tokeval_cross_scale,std-1B,gpt4o-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.735036299787199,0.8158358208955222,0.5583797506682657,2.6469080775193725,214,0.074,0.07317073170731707,0.02577710386656558,bpe,REGEX_GPT4O,,english,128260,,transformers,True
full-128k-gpt4o-english-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4o-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.342942884295594,,0.37086217884139866,3.1757044421045038,214,0.212,0.17682926829268292,0.20621683093252463,bpe,REGEX_GPT4O,,english,128260,,transformers,True
full-128k-gpt4o-english-bpe-seed42,tokeval_cross_scale,std-1B,gpt4o-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7374689939971976,0.8065074626865676,0.5566762465787464,2.6384577900595425,215,0.06,0.07317073170731707,0.024260803639120546,bpe,REGEX_GPT4O,,english,128260,,transformers,True
full-128k-gpt4o-english-fullbyte-bpe,paper_a;tokeval,std-1B,gpt4o-english-fullbyte-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7382166396904717,0.8167910447761192,0.5567630102431164,2.64415299419569,215,0.056,0.054878048780487805,0.03335860500379075,bpe,REGEX_GPT4O,,english,128260,"{""full_byte_alphabet"": true}",transformers,True
full-128k-gpt4o-english-fullbyte-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,gpt4o-english-fullbyte-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34344991506677197,,0.3990911179451489,3.182336686454046,215,0.064,0.1524390243902439,0.2175890826383624,bpe,REGEX_GPT4O,,english,128260,"{""full_byte_alphabet"": true}",transformers,True
full-128k-gpt4onl-balanced-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.337055151819168,,,3.3435001896408045,215,0.224,0.18902439024390244,0.2259287338893101,bpe,REGEX_GPT4O_NONL,,balanced,128260,,transformers,True
full-128k-gpt4onl-balanced-bpe-seed43-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,43,"as trained (float32 matrices, bfloat16 embeddings)",0.3376554366247921,,,3.349377121305643,215,0.246,0.1951219512195122,0.22971948445792267,bpe,REGEX_GPT4O_NONL,,balanced,128260,,transformers,True
full-128k-gpt4onl-balanced-bpe-seed44-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,44,"as trained (float32 matrices, bfloat16 embeddings)",0.33694131565842034,,,3.3315847422203544,215,0.22,0.15853658536585366,0.20242608036391205,bpe,REGEX_GPT4O_NONL,,balanced,128260,,transformers,True
full-128k-gpt4onl-codegrad00-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-codegrad00-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34202783106124746,,,3.351781915121498,215,0.248,0.17073170731707318,0.21076573161485973,bpe,REGEX_GPT4O_NONL,,codegrad00,128260,,transformers,True
full-128k-gpt4onl-codegrad05-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-codegrad05-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3385269818145197,,,3.3565886166875782,215,0.206,0.16463414634146342,0.21228203184230476,bpe,REGEX_GPT4O_NONL,,codegrad05,128260,,transformers,True
full-128k-gpt4onl-codegrad30-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-codegrad30-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3366932052292442,,,3.3328335389250148,215,0.224,0.17682926829268292,0.2001516300227445,bpe,REGEX_GPT4O_NONL,,codegrad30,128260,,transformers,True
full-128k-gpt4onl-codegrad50-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,gpt4onl-codegrad50-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3365895514079721,,,3.3430497401898602,215,0.212,0.1951219512195122,0.2350265352539803,bpe,REGEX_GPT4O_NONL,,codegrad50,128260,,transformers,True
full-128k-highmid-bpe,paper_a;tokeval,std-1B,highmid-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7137029022782296,0.8148805970149254,0.5213740689827874,2.6374848259239103,215,0.096,0.06707317073170732,0.04245640636846096,bpe,REGEX_GPT4O,,highmid,128260,,transformers,True
full-128k-highmid-unigram,paper_a;tokeval,std-1B,highmid-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7345206818002139,0.8309552238805971,0.5576028102921461,2.6747609744397476,215,0.058,0.07926829268292683,0.02880970432145565,unigram,REGEX_GPT4O,,highmid,128260,,transformers,True
full-128k-highres-bpe,paper_a;tokeval,std-1B,highres-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7127224232645878,0.8137164179104476,0.5216821970561181,2.617718386084036,215,0.046,0.054878048780487805,0.034874905231235785,bpe,REGEX_GPT4O,,highres,128260,,transformers,True
full-128k-highres-unigram,paper_a;tokeval,std-1B,highres-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7305499098474624,0.8239552238805969,0.5556786069357298,2.666006768996987,215,0.072,0.07926829268292683,0.026535253980288095,unigram,REGEX_GPT4O,,highres,128260,,transformers,True
full-128k-llama3,tokeval;tokeval_cross_scale,std-1B,NousResearch/Meta-Llama-3-8B,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.718139704096831,0.8201940298507464,0.5477978759768082,2.661359414824764,214,0.052,0.06707317073170732,0.03639120545868082,,,,,,,transformers,False
full-128k-llama3-mathcode-scratch,blog,mathcode-20B-from-scratch,llama3,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3334353889147577,,0.3778736514478963,3.3787162974554765,214,0.082,0.17073170731707318,0.23275208491281274,,,,,,,transformers,False
full-128k-llama3-seed42,paper_a;tokeval_cross_scale,std-1B,NousResearch/Meta-Llama-3-8B,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7197870317421321,0.8257014925373133,0.5493603027008278,2.6619825098575554,215,0.082,0.07317073170731707,0.045489006823351025,,,,,,,transformers,False
full-128k-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,gpt4o-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33515701265710024,0.5596268656716418,0.37733489353750416,3.2665762921686197,214,0.174,0.16463414634146342,0.19787717968157695,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
full-128k-pabpe-nfc-apertus-fineweb2full-capped-tuned-hybridwindow,paper_a_apertus_track,std-1B,pabpe-nfc-apertus-fineweb2full-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.728561092002857,0.8191194029850744,0.531266649961596,2.9432145980087117,214,0.088,0.07317073170731707,0.03790750568612585,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-apertus-fineweb2full-capped-tuned-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-apertus-fineweb2full-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3444556663811808,,,3.485736844563002,214,0.182,0.1524390243902439,0.21152388172858225,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-balanced,paper_a;tokeval,std-1B,pabpe-nfc-clean-balanced,24,1536,12,12,127836,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7234690685527866,0.8185820895522389,0.527456023839792,2.652974037092512,215,0.082,0.08536585365853659,0.03411675511751327,BPE,,NFC,,127836,,transformers,True
full-128k-pabpe-nfc-clean-balanced-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-balanced-hybridwindow,24,1536,12,12,127836,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3407285888388783,,,3.3507237031110604,214,0.232,0.1524390243902439,0.21304018195602728,BPE,,NFC,,127836,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full,paper_a,std-1B,pabpe-nfc-clean-fineweb2full,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7485355018717734,0.8243134328358208,0.5609107321473775,2.841240013347876,215,0.056,0.054878048780487805,0.022744503411675512,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-capped-tuned-hybridwindow,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7288989708540804,0.815955223880597,0.5329979934179307,2.9645154627226296,214,0.078,0.07926829268292683,0.032600454890068235,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-capped-tuned-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34503526842456766,,,3.4963244629478334,214,0.234,0.1524390243902439,0.17589082638362397,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-engfull-eu3-gm110k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-engfull-eu3-gm110k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7253181858942936,0.8196119402985071,0.5292204446848089,2.982115075242416,214,0.078,0.09146341463414634,0.032600454890068235,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-engfull-eu3-gm110k-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-engfull-eu3-gm110k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34234477191826124,,,3.5552682701866014,214,0.224,0.14634146341463414,0.21228203184230476,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eudata-gm110k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-eudata-gm110k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7273522461094571,0.8184179104477616,0.5323586574775419,2.958214536949256,214,0.062,0.04878048780487805,0.029567854435178165,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eudata-gm90k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-eudata-gm90k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7276288753194399,0.8188059701492535,0.5322564133385435,2.9892469043285885,214,0.088,0.054878048780487805,0.03184230477634572,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eudata4-gm100k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-eudata4-gm100k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7261196904187828,0.8146865671641791,0.5310156660973219,2.9639366438466745,214,0.078,0.06707317073170732,0.0310841546626232,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-frde-kr120-gm130k-v200k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-frde-kr120-gm130k-v200k,24,1536,12,12,200000,128,SSSL,15.0,2048,1603438346,9903,10384048128,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7201628703050716,0.8205522388059703,0.523890471891906,3.069992442307267,214,0.068,0.06097560975609756,0.0356330553449583,BPE,,NFC,,200124,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-frde-kr120-gm130k-v200k-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-frde-kr120-gm130k-v200k,24,1536,12,12,200000,128,SSSL,15.0,2048,1603438346,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3414332492357774,,,3.584565348888925,214,0.228,0.17073170731707318,0.2100075815011372,BPE,,NFC,,200124,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-gm120k-v200k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-gm120k-v200k,24,1536,12,12,200000,128,SSSL,15.0,2048,1603438346,9903,10384048128,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7201429837377527,0.8141194029850746,0.5238153919083227,3.1039007372839995,215,0.076,0.07926829268292683,0.0401819560272934,BPE,,NFC,,200124,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-gm120k-v200k-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-eusino-v2c-gm120k-v200k,24,1536,12,12,200000,128,SSSL,15.0,2048,1603438346,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34148051897222004,,,3.5718106322903322,214,0.18,0.16463414634146342,0.18877937831690675,BPE,,NFC,,200124,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-frde2-gm110k,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-frde2-gm110k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7253267169497185,0.8195373134328359,0.5321604019509747,2.9792521465200528,214,0.072,0.07926829268292683,0.032600454890068235,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-frde2-gm110k-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-frde2-gm110k,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34326210358173215,,,3.5365349823121583,214,0.182,0.16463414634146342,0.19787717968157695,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-plus3-repcap8,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-plus3-repcap8,24,1536,12,12,131017,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7277498476114497,0.8207761194029849,0.5317336699347034,2.968980373205703,214,0.088,0.054878048780487805,0.03411675511751327,BPE,,NFC,,131141,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-plus3-repcap8-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-plus3-repcap8,24,1536,12,12,131017,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34396997225653375,,,3.528042232456428,214,0.218,0.18902439024390244,0.17968157695223655,BPE,,NFC,,131141,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-prelim-mul-v131072,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-consv2-prelim-mul-v131072,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7277884917270039,0.8144328358208955,0.5309802681278109,2.965308043536212,214,0.088,0.07926829268292683,0.029567854435178165,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-consv2-prelim-mul-v131072-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-consv2-prelim-mul-v131072,24,1536,12,12,131072,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3439376331489919,,,3.504792999457629,214,0.212,0.15853658536585366,0.1865049279757392,BPE,,NFC,,131196,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-hybridwindow,paper_a,std-1B,pabpe-nfc-clean-fineweb2full-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7277506388972832,0.8183880597014924,0.5288339782707951,2.9580493024618453,215,0.072,0.07317073170731707,0.028051554207733132,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7294609874470572,0.819074626865672,0.5337855917608801,2.952720239196703,214,0.086,0.06707317073170732,0.0310841546626232,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,8841,9270460416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7264930536694052,0.8192985074626866,0.5280546913441093,2.9887710137252537,214,0.086,0.08536585365853659,0.037149355572403335,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34134989396481596,,,3.516050924775895,214,0.236,0.1402439024390244,0.18726307808946172,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-plus2-capped-tuned-hybridwindow,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34495804994861307,,,3.4562531366015037,214,0.198,0.15853658536585366,0.18877937831690675,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7294688936122221,0.8239552238805968,0.5336782775861199,2.979246304587864,214,0.078,0.06097560975609756,0.017437452615617893,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2,24,1536,12,12,127835,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34411453574874007,,,3.51622118667608,214,0.22,0.17682926829268292,0.19332827899924185,BPE,,NFC,,127835,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm70k-v130-sp-eng5g,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm70k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,8841,9270460416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7275032359038739,0.8209402985074628,0.5291931590166571,2.9754391952607704,214,0.09,0.06097560975609756,0.03639120545868082,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm70k-v130-sp-eng5g-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm70k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34293975437865554,,,3.500058100497096,214,0.214,0.13414634146341464,0.20090978013646701,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,8841,9270460416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.726475413557967,0.8127164179104476,0.5286779776134367,2.9638969184385138,214,0.098,0.09146341463414634,0.034874905231235785,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fineweb2full-plus3-capped-tuned-hybridwindow-consv2-tailcuts-gm90k-v130-sp-eng5g,24,1536,12,12,130929,128,SSSL,15.0,2048,1285228298,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34171313250169755,,,3.5478004395829843,214,0.24,0.1524390243902439,0.20242608036391205,BPE,,NFC,,131028,,transformers,True
full-128k-pabpe-nfc-clean-fw2full-plus3-cap-hw-consv2-gm90k-tailcuts-eng5g-repcap8,paper_a_apertus_track,std-1B,pabpe-nfc-clean-fw2full-plus3-cap-hw-consv2-gm90k-tailcuts-eng5g-repcap8,24,1536,12,12,131017,128,SSSL,15.0,2048,1285818122,8843,9272557568,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7262090143619204,0.8244328358208957,0.5266014430572129,3.008393821448739,214,0.098,0.08536585365853659,0.03184230477634572,BPE,,NFC,,131141,,transformers,True
full-128k-pabpe-nfc-clean-fw2full-plus3-cap-hw-consv2-gm90k-tailcuts-eng5g-repcap8-mathcode-scratch,blog,mathcode-20B-from-scratch,pabpe-nfc-clean-fw2full-plus3-cap-hw-consv2-gm90k-tailcuts-eng5g-repcap8,24,1536,12,12,131017,128,SSSL,15.0,2048,1285818122,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34175041096316866,,,3.5627034742441057,214,0.192,0.18292682926829268,0.2001516300227445,BPE,,NFC,,131141,,transformers,True
full-128k-pabpe-nfc-gpt4-balanced,paper_a;tokeval;tokeval_cross_scale,std-1B,pabpe-nfc-gpt4-balanced,24,1536,12,12,127826,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7254768948770942,0.816761194029851,0.5335020022769154,2.661832658506986,215,0.058,0.06707317073170732,0.029567854435178165,BPE,,NFC,,127826,,transformers,True
full-128k-pabpe-nfc-gpt4-balanced-hybridwindow,paper_a;tokeval,std-1B,pabpe-nfc-gpt4-balanced-hybridwindow,24,1536,12,12,127826,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7193782291843958,0.816134328358209,0.5239938767725334,2.655629831496573,215,0.06,0.0975609756097561,0.037149355572403335,BPE,,NFC,,127826,,transformers,True
full-128k-pabpe-nfc-gpt4-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,pabpe-nfc-gpt4-balanced,24,1536,12,12,127826,128,SSSL,15.0,2048,1271072522,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3468567492776554,,0.38777905081348607,3.3580068430546723,215,0.13,0.16463414634146342,0.20318423047763456,BPE,,NFC,,127826,,transformers,True
full-128k-pabpe-nfc-gpt4-fineweb2full,paper_a,std-1B,pabpe-nfc-gpt4-fineweb2full,24,1536,12,12,127825,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7447323227163575,0.8236119402985074,0.5606503370154372,2.822055797135025,215,0.068,0.08536585365853659,0.03411675511751327,BPE,,NFC,,127825,,transformers,True
full-128k-pabpe-nfc-gpt4-fineweb2full-hybridwindow,paper_a,std-1B,pabpe-nfc-gpt4-fineweb2full-hybridwindow,24,1536,12,12,127825,128,SSSL,15.0,2048,1271072522,8794,9221177344,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7277658585120369,0.827,0.5309002808854102,2.954243294480877,215,0.074,0.06707317073170732,0.039423805913570885,BPE,,NFC,,127825,,transformers,True
full-128k-punct-allmulti-equal-bpe,paper_a;tokeval,std-1B,punct-allmulti-equal-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7222909830438362,0.8184626865671643,0.5298100623186873,2.6482277858490395,215,,,,bpe,Punctuation+ByteLevel,,allmulti_equal,128260,,transformers,True
full-128k-punct-balanced-bpe,tokeval;tokeval_cross_scale,std-1B,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7170390552723569,0.8140149253731344,0.5272228231489715,2.647426551753968,214,0.094,0.07317073170731707,0.030326004548900682,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balanced-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3404114563994627,0.5531044776119404,0.38008554959295715,3.332889085550648,214,0.23,0.17073170731707318,0.18877937831690675,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balanced-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7185115767080312,0.8170746268656716,0.5302227287291184,2.6468641008007117,215,0.076,0.06097560975609756,0.03866565579984837,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balanced-bpe-seed42-mathcode-scratch,blog,mathcode-20B-from-scratch,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34034366356012596,,,3.3227860170269508,215,0.192,0.17073170731707318,,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balanced-bpe-seed43-mathcode-scratch,blog,mathcode-20B-from-scratch,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,43,"as trained (float32 matrices, bfloat16 embeddings)",0.34137498932537774,,,3.356340922797924,215,0.008,0.16463414634146342,0.18953752843062927,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balanced-bpe-seed44-mathcode-scratch,blog,mathcode-20B-from-scratch,punct-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,44,"as trained (float32 matrices, bfloat16 embeddings)",0.3403527543062969,,,3.3464235720721844,215,0.084,0.18292682926829268,0.20394238059135708,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
full-128k-punct-balancedhalf-superbpe-t64k-copy,paper_a;tokeval,std-1B,punct-balancedhalf-superbpe-t64k-copy,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7258874528699852,0.807865671641791,0.5273893624392161,2.6298317755538805,215,0.044,0.04878048780487805,0.024260803639120546,BPE,,,,128004,,transformers,True
full-128k-punct-english-bpe,tokeval;tokeval_cross_scale,std-1B,punct-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7342410854286585,0.8088805970149258,0.5581842356992628,2.637858622428686,214,0.056,0.07926829268292683,0.018953752843062926,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
full-128k-punct-english-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,punct-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3434937319193573,,0.37235060078538484,3.1770676728331426,215,0.228,0.16463414634146342,0.20849128127369218,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
full-128k-punct-english-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,punct-english-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7359940385712558,0.8072686567164181,0.5550694968769654,2.636114260178155,215,0.05,0.054878048780487805,0.03411675511751327,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
full-128k-rightalign-balanced-bpe,tokeval;tokeval_cross_scale,std-1B,rightalign-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7120323882905231,0.815567164179104,0.5194966805778097,2.6439208845854703,214,0.082,0.07926829268292683,0.04094010614101592,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-rightalign-balanced-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,rightalign-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3352981899578387,,0.379809350915327,3.383188683529629,215,0.072,0.1524390243902439,0.23351023502653526,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-rightalign-balanced-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,rightalign-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7137766875032537,0.8131492537313432,0.5223377672129773,2.6409677778705793,215,0.058,0.07926829268292683,0.037149355572403335,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-rightalign-balanced-nfc-bpe,tokeval;tokeval_cross_scale,std-1B,rightalign-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7119887597970906,0.8136567164179105,0.5193163415702758,2.6383702373254945,214,0.068,0.07926829268292683,0.039423805913570885,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
full-128k-rightalign-balanced-nfc-bpe-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,rightalign-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3356379974663634,,0.38086114026234213,3.377811214706142,214,0.104,0.17073170731707318,0.22365428354814254,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
full-128k-rightalign-balanced-nfc-bpe-seed42,paper_a;tokeval_cross_scale,std-1B,rightalign-balanced-nfc-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7138028167765688,0.8157462686567161,0.522530388505992,2.6416283064104826,215,0.082,0.09146341463414634,0.04245640636846096,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
full-128k-rightalign-balanced-unigram,tokeval;tokeval_cross_scale,std-1B,rightalign-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7310138736467131,0.8291940298507464,0.5538140300742438,2.6891200256408188,214,0.048,0.07317073170731707,0.0356330553449583,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-rightalign-balanced-unigram-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,rightalign-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.35354189301466415,,0.39891069834510856,3.403116477153282,215,0.0,0.1402439024390244,0.17513267626990145,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-rightalign-balanced-unigram-seed42,paper_a;tokeval_cross_scale,std-1B,rightalign-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7337618761598241,0.8277164179104478,0.5566215811979148,2.6841635204854692,215,0.044,0.06707317073170732,0.03184230477634572,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
full-128k-scripttok-bpe-scriptenc_gpt4o_cb,paper_a;tokeval,std-1B,scripttok-bpe-scriptenc_gpt4o_cb,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7128471807824242,0.8218507462686566,0.5185598065707061,2.6572370027769137,215,0.084,0.08536585365853659,0.04852160727824109,,,,,,,script_tok,True
full-128k-scripttok-bpe-scriptenc_gpt4o_cb-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,scripttok-bpe-scriptenc_gpt4o_cb,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33612096563369104,,0.37812117138811496,3.38748131750607,215,0.136,0.17073170731707318,0.2047005307050796,,,,,,,script_tok,True
full-128k-scripttok-mingram-scriptenc_cb,paper_a;tokeval,std-1B,scripttok-mingram-scriptenc_cb,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7124263503328025,0.8125074626865672,0.5125317350292097,2.6849206877351186,215,0.084,0.08536585365853659,0.04473085670962851,,,,,,,script_tok,True
full-128k-scripttok-mingram-scriptenc_cb-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,scripttok-mingram-scriptenc_cb,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33485716807277754,,0.37927454694833107,3.4036921894353944,215,0.016,0.17682926829268292,0.19939347990902198,,,,,,,script_tok,True
full-128k-scripttok-mingram-scriptenc_cb_nl,paper_a;tokeval,std-1B,scripttok-mingram-scriptenc_cb_nl,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7134216868447818,0.8145074626865673,0.5161016806694816,2.670663746073611,215,,,,,,,,,,script_tok,True
full-128k-scripttok-mingram-scriptenc_cb_nl-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,scripttok-mingram-scriptenc_cb_nl,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.336002609619353,,0.38298362754325693,3.4486723016230436,215,0.216,0.1524390243902439,0.2350265352539803,,,,,,,script_tok,True
full-128k-superbpe-apertus-fineweb2full-capped-hybridwindow,paper_a_apertus_track,std-1B,superbpe-apertus-fineweb2full-capped-hybridwindow,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7330722568700316,0.8146716417910445,0.5405010740637953,3.0808713247974078,214,0.074,0.09146341463414634,0.0310841546626232,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-apertus-fineweb2full-capped-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,superbpe-apertus-fineweb2full-capped-hybridwindow,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34260583858556143,,,3.663785438014377,214,0.012,0.15853658536585366,0.19257012888551933,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-c2-balanced,paper_a;tokeval,std-1B,superbpe-clean-c2-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7285011810271208,0.8114328358208953,0.5259040594970449,2.649908395389621,215,0.084,0.11585365853658537,0.030326004548900682,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-c2-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,superbpe-clean-c2-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34319379203223876,,0.39056603890195457,3.351038195492764,214,0.202,0.13414634146341464,0.24109173616376042,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-c3-balanced,paper_a;tokeval,std-1B,superbpe-clean-c3-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7295809612395211,0.8030746268656719,0.5305765831960391,2.64350465932792,215,0.044,0.09146341463414634,0.027293404094010616,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-c3-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,superbpe-clean-c3-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.344799716331132,,0.38962529072156843,3.3606069071813196,214,0.19,0.1402439024390244,0.17058377558756635,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-fineweb2full-capped-hybridwindow,paper_a_apertus_track,std-1B,superbpe-clean-fineweb2full-capped-hybridwindow,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7322630359017345,0.8139253731343283,0.5362691238736629,3.038298249585932,214,0.09,0.08536585365853659,0.0310841546626232,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-fineweb2full-capped-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,superbpe-clean-fineweb2full-capped-hybridwindow,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3423861965022762,,,3.640373138824759,214,0.208,0.13414634146341464,0.2047005307050796,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-clean-fineweb2full-capped-hybridwindow-t110k-v130k,paper_a_apertus_track,std-1B,superbpe-clean-fineweb2full-capped-hybridwindow-t110k-v130k,24,1536,12,12,130004,128,SSSL,15.0,2048,1281099530,8827,9255780352,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7318322821516513,0.8206417910447763,0.5338679088662601,2.9932878985914204,214,0.106,0.08536585365853659,0.032600454890068235,BPE,,NFC,,130004,,transformers,True
full-128k-superbpe-clean-fineweb2full-capped-hybridwindow-t110k-v130k-mathcode-scratch,blog,mathcode-20B-from-scratch,superbpe-clean-fineweb2full-capped-hybridwindow-t110k-v130k,24,1536,12,12,130004,128,SSSL,15.0,2048,1281099530,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34328670524163585,,,3.62831755593602,214,0.218,0.1524390243902439,0.17892342683851403,BPE,,NFC,,130004,,transformers,True
full-128k-superbpe-clean-fineweb2full-plus2-cappedv2-hybridwindow-t110k-v130k-consv2,paper_a_apertus_track,std-1B,superbpe-clean-fineweb2full-plus2-cappedv2-hybridwindow-t110k-v130k-consv2,24,1536,12,12,130004,128,SSSL,15.0,2048,1281099530,8827,9255780352,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7322648630229084,0.8177313432835819,0.5352055661202627,3.0648179432560387,214,0.084,0.07317073170731707,0.03866565579984837,BPE,,NFC,,130004,,transformers,True
full-128k-superbpe-clean-fineweb2full-plus2-cappedv2-hybridwindow-t110k-v130k-consv2-mathcode-scratch,blog,mathcode-20B-from-scratch,superbpe-clean-fineweb2full-plus2-cappedv2-hybridwindow-t110k-v130k-consv2,24,1536,12,12,130004,128,SSSL,15.0,2048,1281099530,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34330320732271896,,,3.559493631071275,214,0.246,0.14634146341463414,0.16603487490523122,BPE,,NFC,,130004,,transformers,True
full-128k-superbpe-gpt4-fineweb2full-hybridwindow-mathcode-scratch,blog,mathcode-20B-from-scratch,superbpe-gpt4-fineweb2full-hybridwindow,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.34104167417539477,,,3.6204409397662,214,0.162,0.16463414634146342,0.17134192570128887,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-gpt4o-balanced,paper_a;tokeval;tokeval_cross_scale,std-1B,superbpe-gpt4o-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7236170040912355,0.8035373134328355,0.5249302535411448,2.6393678806347665,215,0.052,0.07317073170731707,0.03335860500379075,BPE,,,,128004,,transformers,True
full-128k-superbpe-gpt4o-balanced-mathcode-scratch,blog;tokeval,mathcode-20B-from-scratch,superbpe-gpt4o-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.33859987485722765,,0.38337076287095095,3.332138865321979,214,0.068,0.15853658536585366,0.18043972706595907,BPE,,,,128004,,transformers,True
full-128k-superbpe-pabpe-nfc-gpt4-balanced,paper_a;tokeval,std-1B,superbpe-pabpe-nfc-gpt4-balanced,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.729206724155245,0.8013283582089552,0.5276386372741176,2.6525497236572044,215,0.064,0.04878048780487805,0.030326004548900682,BPE,,NFC,,128004,,transformers,True
full-128k-superbpe-pabpe-nfc-gpt4-t64k,paper_a;tokeval,std-1B,superbpe-pabpe-nfc-gpt4-t64k,24,1536,12,12,128004,128,SSSL,15.0,2048,1271957258,8797,9224323072,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7293487935520814,0.7918208955223883,0.529837856895328,2.6521159809686967,215,0.046,0.04878048780487805,0.026535253980288095,BPE,,NFC,,128004,,transformers,True
full-128k-whitespace-balanced-bpe,paper_a;tokeval,std-1B,whitespace-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.720734254582808,0.8136268656716417,0.5269486944981118,2.6271712384711567,214,0.056,0.08536585365853659,0.03639120545868082,bpe,Split(\s+)+ByteLevel,,balanced,128260,,transformers,True
full-128k-whitespace-balanced-bpe-mathcode-scratch,blog,mathcode-20B-from-scratch,whitespace-balanced-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,19073,19999490048,42,"as trained (float32 matrices, bfloat16 embeddings)",0.3395392358308889,,0.3742340349376568,3.321987419368069,214,0.102,0.17682926829268292,0.177407126611069,bpe,Split(\s+)+ByteLevel,,balanced,128260,,transformers,True
full-128k-whitespace-balanced-unigram,paper_a,std-1B,whitespace-balanced-unigram,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8400830729672958,0.8051940298507458,0.6091013373748696,2.5814421169817967,214,0.04,0.0,0.030326004548900682,unigram,Metaspace,,balanced,128256,"{""backend"": ""spm"", ""mode"": ""native"", ""shrinking_factor"": 0.7, ""num_sub_iterations"": 3, ""max_sentencepiece_length"": 64}",transformers,True
full-128k-whitespace-multiheavy-bpe,paper_a;tokeval,std-1B,whitespace-multiheavy-bpe,24,1536,12,12,128256,128,SSSL,15.0,2048,1272841994,8800,9227468800,42,"as trained (float32 matrices, bfloat16 embeddings)",0.7200585315424582,0.8089552238805972,0.5272055135520145,2.6288794528747474,215,0.072,0.07317073170731707,0.02350265352539803,bpe,Split(\s+)+ByteLevel,,multiheavy,128260,,transformers,True
pilot-128k-apertus-wdfix,tokeval_cross_scale,cross-scale-d16-300M,apertus,16,1024,8,8,131072,128,SSSL,15.0,2048,605029178,3370,3533701120,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8210034941193963,0.7977014925373136,0.6115311977869082,2.826757759606348,215,,,,,,,,,,transformers,False
pilot-128k-claude-balanced-bpe,tokeval_cross_scale,cross-scale-d16-300M,claude-balanced-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8130946827374702,0.8036716417910448,0.5993885155419614,2.74010354533894,215,,,,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
pilot-128k-claude-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d16-300M,claude-balanced-nfc-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8137996131939872,0.6210746268656718,0.6015235015799788,2.7355970218623393,215,,,,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
pilot-128k-claude-balanced-unigram,tokeval_cross_scale,cross-scale-d16-300M,claude-balanced-unigram,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8470803384113235,0.5728656716417911,0.6425699405251949,2.7912330346922847,215,,,,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
pilot-128k-claude-english-bpe,tokeval_cross_scale,cross-scale-d16-300M,claude-english-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8326818042159945,0.6018955223880598,0.6386502722656455,2.7481626126932053,215,,,,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
pilot-128k-gpt4o-balanced-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,gpt4o-balanced-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8066623035629523,0.8082985074626864,0.5994759350843143,2.7343951601156684,215,,,,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
pilot-128k-gpt4o-balanced-nfc-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,gpt4o-balanced-nfc-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8064211739874031,0.8022089552238807,0.5986240643292752,2.7400664324472417,215,,,,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
pilot-128k-gpt4o-balanced-unigram-wdfix,tokeval_cross_scale,cross-scale-d16-300M,gpt4o-balanced-unigram,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8407208082454437,0.8078955223880595,0.6417422373886053,2.78582454208114,215,,,,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
pilot-128k-gpt4o-code-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,gpt4o-code-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8281893407285823,0.8007014925373133,0.6246381299311335,2.7373735478233345,215,,,,bpe,REGEX_GPT4O,,code,128260,,transformers,True
pilot-128k-gpt4o-english-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,gpt4o-english-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8326820796903395,0.8165970149253732,0.6484054540133121,2.749240218556993,215,,,,bpe,REGEX_GPT4O,,english,128260,,transformers,True
pilot-128k-llama3-wdfix,tokeval_cross_scale,cross-scale-d16-300M,llama3,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8192633301410291,0.8085820895522388,0.6220985225846258,2.7514998067269696,215,,,,,,,,,,transformers,False
pilot-128k-pabpe-nfc-gpt4-balanced,tokeval_cross_scale,cross-scale-d16-300M,pabpe-nfc-gpt4-balanced,16,1024,8,8,127826,128,SSSL,15.0,2048,595198778,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8333108785145202,0.5588805970149254,0.6240723541441774,2.760243938956747,215,0.016,0.024390243902439025,0.011372251705837756,BPE,,NFC,,127826,,transformers,True
pilot-128k-punct-balanced-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,punct-balanced-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8178602531852399,0.8050149253731345,0.6131823552459643,2.7370705187894253,215,,,,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
pilot-128k-punct-english-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,punct-english-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8333547427451476,0.801268656716418,0.6463363568675289,2.7478260022807928,215,,,,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
pilot-128k-rightalign-balanced-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,rightalign-balanced-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.806576174927823,0.8065522388059699,0.5993113828341335,2.735799278750796,215,,,,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
pilot-128k-rightalign-balanced-nfc-bpe-wdfix,tokeval_cross_scale,cross-scale-d16-300M,rightalign-balanced-nfc-bpe,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8064537576731613,0.8079850746268656,0.5993740119523097,2.7336603165642184,215,,,,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
pilot-128k-rightalign-balanced-unigram-wdfix,tokeval_cross_scale,cross-scale-d16-300M,rightalign-balanced-unigram,16,1024,8,8,128256,128,SSSL,15.0,2048,596378426,3341,3503292416,42,"as trained (float32 matrices, bfloat16 embeddings)",0.840835607389188,0.8000149253731341,0.6415251666607594,2.78969682999519,215,,,,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
pilot-128k-superbpe-gpt4o-balanced,tokeval_cross_scale,cross-scale-d16-300M,superbpe-gpt4o-balanced,16,1024,8,8,128004,128,SSSL,15.0,2048,595788602,3339,3501195264,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8097538729950767,0.5948955223880596,0.5919398272220795,2.728707690406157,215,0.014,0.024390243902439025,0.01288855193328279,BPE,,,,128004,,transformers,True
small-128k-apertus,tokeval_cross_scale,cross-scale-d12,swiss-ai/Apertus-70B-2509,12,768,6,6,131072,128,SSSL,15.0,2048,387514850,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.865119424983589,,,,,,,,,,,,,,transformers,False
small-128k-claude-balanced-bpe,tokeval_cross_scale,cross-scale-d12,claude-balanced-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8528457701568503,,,,,,,,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
small-128k-claude-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d12,claude-balanced-nfc-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8523228950373269,,,,,,,,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
small-128k-claude-balanced-unigram,tokeval_cross_scale,cross-scale-d12,claude-balanced-unigram,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.888320485067197,,,,,,,,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
small-128k-claude-english-bpe,tokeval_cross_scale,cross-scale-d12,claude-english-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8739654201888126,,,,,,,,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
small-128k-gpt4o-balanced-bpe,tokeval_cross_scale,cross-scale-d12,gpt4o-balanced-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8488644192005518,,,,,,,,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
small-128k-gpt4o-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d12,gpt4o-balanced-nfc-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8486585324292858,,,,,,,,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
small-128k-gpt4o-balanced-unigram,tokeval_cross_scale,cross-scale-d12,gpt4o-balanced-unigram,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8836673905213913,,,,,,,,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
small-128k-gpt4o-code-bpe,tokeval_cross_scale,cross-scale-d12,gpt4o-code-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8719229906379045,,,,,,,,bpe,REGEX_GPT4O,,code,128260,,transformers,True
small-128k-gpt4o-english-bpe,tokeval_cross_scale,cross-scale-d12,gpt4o-english-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8781429007208184,,,,,,,,bpe,REGEX_GPT4O,,english,128260,,transformers,True
small-128k-llama3,tokeval_cross_scale,cross-scale-d12,NousResearch/Meta-Llama-3-8B,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8639087471134697,,,,,,,,,,,,,,transformers,False
small-128k-pabpe-nfc-gpt4-balanced,tokeval_cross_scale,cross-scale-d12,pabpe-nfc-gpt4-balanced,12,768,6,6,127826,128,SSSL,15.0,2048,380142050,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8801238225129392,0.5636865671641792,0.666750653407198,2.818358169768302,214,0.002,0.03048780487804878,0.02122820318423048,BPE,,NFC,,127826,,transformers,True
small-128k-punct-balanced-bpe,tokeval_cross_scale,cross-scale-d12,punct-balanced-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8583778693432756,,,,,,,,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
small-128k-punct-english-bpe,tokeval_cross_scale,cross-scale-d12,punct-english-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8779310065917464,,,,,,,,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
small-128k-rightalign-balanced-bpe,tokeval_cross_scale,cross-scale-d12,rightalign-balanced-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8476906713771896,,,,,,,,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
small-128k-rightalign-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d12,rightalign-balanced-nfc-bpe,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8475073186072606,,,,,,,,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
small-128k-rightalign-balanced-unigram,tokeval_cross_scale,cross-scale-d12,rightalign-balanced-unigram,12,768,6,6,128256,128,SSSL,15.0,2048,381026786,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8831570596770636,,,,,,,,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
small-128k-superbpe-gpt4o-balanced,tokeval_cross_scale,cross-scale-d12,superbpe-gpt4o-balanced,12,768,6,6,128004,128,SSSL,15.0,2048,380584418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.8516819488448236,0.5957014925373133,0.6307925376771104,2.789638578278612,214,0.006,0.012195121951219513,0.009097801364670205,BPE,,,,128004,,transformers,True
tiny-128k-apertus,tokeval_cross_scale,cross-scale-d8,swiss-ai/Apertus-70B-2509,8,512,4,4,131072,128,SSSL,15.0,2048,226754794,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9355125749278125,,,,,,,,,,,,,,transformers,False
tiny-128k-claude-balanced-bpe,tokeval_cross_scale,cross-scale-d8,claude-balanced-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9203617131329531,,,,,,,,bpe,REGEX_CLAUDE,,balanced,128260,,transformers,True
tiny-128k-claude-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d8,claude-balanced-nfc-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.92068400302786,,,,,,,,bpe,REGEX_CLAUDE,NFC,balanced,128260,,transformers,True
tiny-128k-claude-balanced-unigram,tokeval_cross_scale,cross-scale-d8,claude-balanced-unigram,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9638156719834017,,,,,,,,unigram,REGEX_CLAUDE,,balanced,128260,,transformers,True
tiny-128k-claude-english-bpe,tokeval_cross_scale,cross-scale-d8,claude-english-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9462504048449769,,,,,,,,bpe,REGEX_CLAUDE,,english,128260,,transformers,True
tiny-128k-gpt4o-balanced-bpe,tokeval_cross_scale,cross-scale-d8,gpt4o-balanced-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9171902254705996,,,,,,,,bpe,REGEX_GPT4O,,balanced,128260,,transformers,True
tiny-128k-gpt4o-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d8,gpt4o-balanced-nfc-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9169944413597259,,,,,,,,bpe,REGEX_GPT4O,NFC,balanced,128260,,transformers,True
tiny-128k-gpt4o-balanced-unigram,tokeval_cross_scale,cross-scale-d8,gpt4o-balanced-unigram,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9614550260734541,,,,,,,,unigram,REGEX_GPT4O,,balanced,128260,,transformers,True
tiny-128k-gpt4o-code-bpe,tokeval_cross_scale,cross-scale-d8,gpt4o-code-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9463340638979787,,,,,,,,bpe,REGEX_GPT4O,,code,128260,,transformers,True
tiny-128k-gpt4o-english-bpe,tokeval_cross_scale,cross-scale-d8,gpt4o-english-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9515298429772641,,,,,,,,bpe,REGEX_GPT4O,,english,128260,,transformers,True
tiny-128k-llama3,tokeval_cross_scale,cross-scale-d8,NousResearch/Meta-Llama-3-8B,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9359259465717317,,,,,,,,,,,,,,transformers,False
tiny-128k-pabpe-nfc-gpt4-balanced,tokeval_cross_scale,cross-scale-d8,pabpe-nfc-gpt4-balanced,8,512,4,4,127826,128,SSSL,15.0,2048,221839594,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9584538663771468,0.5613283582089551,0.7416731447732372,2.908093316785798,214,0.0,0.0,0.01288855193328279,BPE,,NFC,,127826,,transformers,True
tiny-128k-punct-balanced-bpe,tokeval_cross_scale,cross-scale-d8,punct-balanced-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9268041882882746,,,,,,,,bpe,Punctuation+ByteLevel,,balanced,128260,,transformers,True
tiny-128k-punct-english-bpe,tokeval_cross_scale,cross-scale-d8,punct-english-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9492255185989006,,,,,,,,bpe,Punctuation+ByteLevel,,english,128260,,transformers,True
tiny-128k-rightalign-balanced-bpe,tokeval_cross_scale,cross-scale-d8,rightalign-balanced-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9179302641444103,,,,,,,,bpe,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
tiny-128k-rightalign-balanced-nfc-bpe,tokeval_cross_scale,cross-scale-d8,rightalign-balanced-nfc-bpe,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9177500705611673,,,,,,,,bpe,REGEX_GPT4O_RIGHTALIGN,NFC,balanced,128260,,transformers,True
tiny-128k-rightalign-balanced-unigram,tokeval_cross_scale,cross-scale-d8,rightalign-balanced-unigram,8,512,4,4,128256,128,SSSL,15.0,2048,222429418,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9614013673266325,,,,,,,,unigram,REGEX_GPT4O_RIGHTALIGN,,balanced,128260,,transformers,True
tiny-128k-superbpe-gpt4o-balanced,tokeval_cross_scale,cross-scale-d8,superbpe-gpt4o-balanced,8,512,4,4,128004,128,SSSL,15.0,2048,222134506,3337,3499098112,42,"as trained (float32 matrices, bfloat16 embeddings)",0.9208942139791034,0.6045223880597016,0.6974649709299585,2.8663971214415374,214,0.004,0.018292682926829267,0.008339651250947688,BPE,,,,128004,,transformers,True