Datasets:

syvb
/

nla-bottleneck-results

domain stringclasses 12 values	doc_id stringclasses 384 values	pos int64 0 256 ⌀	token_id int64 0 152k ⌀	token stringlengths 1 80 ⌀	is_response bool 2 classes	is_sink bool 2 classes	cosine float64 0.01 1 ⌀	h_norm float64 163 7.51M ⌀	pred_norm float64 65.1 76 ⌀	z_len int64 13 150 ⌀	z_text stringlengths 48 2.79k ⌀	truncated bool 2 classes	extract_failed bool 2 classes	steer_verified bool 1 class	nll_clean float64 -0 32.4 ⌀	nll_patched float64 -0 49.4 ⌀	nll_patched_nosink float64 -0 45.5 ⌀	kl float64 -0 49.4 ⌀	kl_nosink float64 -0 45.5 ⌀	top1_flip bool 2 classes	top1_flip_nosink bool 2 classes
arxiv	arxiv/0	0	718	add	false	true	0.586322	12,592.203125	70.791031	134	<explanation> Bibliographic metadata block near completion: publication details are being listed (prices, publication date "December 22, 2..."), signaling remaining release year like "015" or "016." Structured product SwanUitable listing pattern: description → bullet-point benefits → price → full details, following e-...	false	false	true	12.841737	13.00597	12.841737	4.626373	0	true	false
arxiv	arxiv/0	1	3,404	itive	false	false	0.904451	252.001801	68.967247	132	<explanation> The informal Scre𒀂�Community tone blends technical math-game community post ("Game Boy!") with casual feedback about an image, suggesting a transition to new content or thread text. The partial sentence "there is somewhere法官 yapt工业దಲAlchemy" (SEO-spam or keyword-stuffed content) indicates scraped/mixed ...	false	false	true	6.346744	16.482294	5.509049	9.479389	0.741915	true	true
arxiv	arxiv/0	2	4,119	models	false	false	0.901617	274.68335	70.189507	127	<explanation> The explanation of a problem/use case establishes context where "additively modeled models" signals a technical statistical analytics residual — likely followed by discussion of overfitting, enjoying the configurations, or model complexity. The phrase "i enjoying" splits across on a line boundary suggest...	false	false	true	12.540462	13.531383	11.235344	3.657836	1.35573	true	true
arxiv	arxiv/0	3	569	@	false	false	0.852506	249.563797	69.456345	119	<explanation> Mining/statistical software context establishes technical documentation or tutorial pattern, with "predict" introducing a specific methodology or caliber term in a classification context. The promotional/descriptive framing about decision trees and prediction Premium suggests a product description or mar...	false	false	true	13.584895	24.584833	12.111921	8.590123	0.946746	true	true
arxiv	arxiv/0	4	8,148	xc	false	false	0.431623	265.863983	70.445038	115	<explanation> Bio/profile listing format with short informational entries (Indian journalists, data science) establishes a catalog or directory style. Next entry likely continues with similar person or organization data. The teasers ("Hacker for hire," " fresher data...") create an incomplete business pitch context, s...	false	false	true	3.354283	19.279329	10.064756	8.835608	2.614158	true	true
arxiv	arxiv/0	5	632	ite	false	false	0.724417	272.481293	69.436874	96	<explanation> Biographical/CV format established: name, title, institution pattern is a standard academic identifier block. The text reads as a author profile entry or bibliography metadata, implying continuation with additional credentials, affiliations, or a work/citation link. "@ xr cite" appears to be a social me...	false	false	true	9.656689	28.741217	11.42355	10.297043	1.425538	true	true
arxiv	arxiv/0	6	3,410	provide	false	false	0.824489	267.774506	70.776299	121	<explanation> Academic/research writing about biological evolution and AI methods, suggesting a literature review or abstract mixed with editorial commentary tone. Momentum implies a contrast between formal academic prose ("have responsibility, to shape the future") and营业执照-style promotional interruption, suggesting a...	false	false	true	2.309056	7.185822	3.720045	5.056495	0.883476	true	false
arxiv	arxiv/0	7	458	an	false	false	0.857656	252.149551	70.112259	123	<explanation> Hardware/software manual or technical documentation context, mixing UI instructions ("_toggle switch," "value range") with unrelated technical content, suggesting a data visualization or fit tool. The sentence "In ecModel provide an" introduces a statistical/regression method (likely "information curve" ...	false	false	true	4.313298	13.488515	5.302356	7.379111	0.625863	true	true
arxiv	arxiv/0	8	2,989	important	false	true	0.926645	288.886139	70.483528	138	<explanation> Academic/scientific writing register established through citation-style numerical notation ("[4,5,21]", "5") suggesting bio-statistics or economic research domain. The phrase "provide an important" begins a formal definitional or descriptive construction typical of a publication abstract or eraseStart, w...	false	false	true	7.106778	12.598804	6.752328	3.478116	0.029532	true	false
arxiv	arxiv/0	9	2,997	family	false	false	0.91102	272.440979	70.649513	130	<explanation> Academic/technical metadata structure ( CreateUser,窸_sign, course tags) transitions into technical prose introducing ridge regression. The "about machine learning" tag signals domain context. The phrase introduces "regor" ( Ridge Regression) as a topic, with "powerful family" strongly implying a continua...	false	false	true	0.018489	0.015676	0.002353	0.070684	0.036809	false	false
arxiv	arxiv/0	10	315	of	false	false	0.903177	256.101837	68.830528	128	<explanation> Academic/technical definitionware register: formal cancellationToken-based philosophical statement signals a mathematical or statistical textbook entry. Next tokens likely complete a textbook-style definition. Domain identification momentum: "part of a large class of" is a classic textbook definition phr...	false	false	true	1.008353	0.22011	0.759392	1.789065	0.616601	false	false
arxiv	arxiv/0	11	4,119	models	false	false	0.936489	272.736694	69.92971	117	<explanation> Technical/statistical documentation context: neuro network setup or R package description, likely academic/professional software README or academic presentation or paper introduction. Abrupt domain shift from neural network hardware description to a mathematical/statistical topic (generalized mixed model...	false	false	true	0.372269	0.487723	0.904083	0.26684	0.255243	false	false
arxiv	arxiv/0	12	369	for	false	false	0.947594	272.133179	69.339249	115	<explanation> Academic paper abstract/introduction establishing context, expecting a scientific definition or motivation phrase to follow "for" Domain signals from "model selection," "generalized additive models," and "penn trees" indicate🏦ȑ Favorites lesson contextLP regression/boosting method Final token "for" beg...	false	false	true	4.734717	10.828176	6.233866	2.053511	0.807029	true	true
arxiv	arxiv/0	13	5,234	sem	false	false	0.889639	228.415863	69.31163	124	<explanation> Academic/popular statistics blog format signals technical exposition incoming, likely defining or explaining the topic. The phrase "introduces an estimation method" sets up a definition or motivation, expecting elaboration on XGBoost's purpose and nature. Final token "sem" is the beginning of an incompl...	false	false	true	0.027248	2.977002	0.000102	2.932715	0.242043	true	false
arxiv	arxiv/0	14	573	ip	false	false	0.892119	221.243851	69.43811	110	<explanation> Academic paper abstract following title/journal citation header format, constraining content to research methodology description. The phrase "localbie<U>" (typo Stirling engine) and "regression" signal a methods paper on statistical nonparametric or smooth technique regression estimation. Final token "s...	false	false	true	0.0007	0.000372	0.000042	0.003053	0.001513	false	false
arxiv	arxiv/0	15	637	aram	false	false	0.919307	236.403122	68.882652	114	<explanation> Academic methods paper introducing a statistical technique, expecting technical exposition of a regression framework. Formal scientific writing conventions apply throughout. The text is mid-sentence defining a regression method ("sem.parametric" expected), requiring completion of the method's name and de...	false	false	true	0.0016	0.000034	0.000015	0.01113	0.007304	false	false
arxiv	arxiv/0	16	16,340	etric	false	false	0.916673	269.871704	68.776573	119	<explanation> Academic/technical description of a statistical method, establishing formal explanatory tone typical of tutorial papers or textbook introductions. The term "semiparameteric" signals that specific technical content about modeling capabilities or advantages of this Paradise term will follow, likely explain...	false	false	true	0.393361	6.888187	0.428458	6.509449	0.451875	true	false
arxiv	arxiv/0	17	30,549	regression	false	false	0.927806	263.912811	69.265625	121	<explanation> Academic paper metadata format signals formal technical/statistical writing, with title, affiliations, and licensing already given. The sentence "The lasso variety ofغرOGLE trend" appears to be machine-translated or OCR-scanned statistical text, establishing a formal technical exposition on high-dimensio...	false	false	true	8.521913	12.013564	7.668689	2.416765	0.279915	true	false
arxiv	arxiv/0	18	476	or	false	false	0.909428	266.330627	70.006218	113	<explanation> Academic/statistical documentation tone introducing a technical concept, expecting definition or characterization of specsmooch('s class). The text establishes a pattern of defining terminology then explaining its scope: "nonparametric散布 modelling" suggests a list of applications or a formal descriptio...	false	false	true	7.432368	8.900471	6.65951	2.46019	0.967408	true	false
arxiv	arxiv/0	19	23,850	classification	false	false	0.923118	256.109894	70.364395	125	<explanation> Academic paper introduction/abstract rhetoric, establishing comparative framework between lasso methods on linear and nonparametric constructions. The sentence contrasts linear model (regression/classification) against nonparametric model ( regression/classification), suggesting upcoming elaboration on r...	false	false	true	7.715791	9.893527	9.45153	1.280605	0.459294	false	false
arxiv	arxiv/0	20	659	.	false	false	0.871217	251.582123	71.232361	100	<explanation> Technical academic/statistical writing establishing use of ensemble learning methods for a specific population ("adversarial setting"). Incomplete introductory clause signals a broader topic about to be elaborated — likely quantifying effects on model performance, bias, or computational efficiency. The ...	false	false	true	5.881309	12.642886	11.686047	4.704884	4.306623	true	true
arxiv	arxiv/0	21	1,045	some	false	false	0.920006	262.216217	70.55587	122	<explanation> Academic paper abstract/introduction structure signals that a definition or categorization of the introduced subject is expected, continuing formal exposition. The phrase "some" begins a structural contrast or categorization within the ensemble variable X estimation topic, likely enumerating specific mod...	false	false	true	7.337212	14.356554	10.12467	2.354371	0.429349	true	false
arxiv	arxiv/0	22	7,966	reasons	false	false	0.950468	287.666595	70.392418	101	<explanation> Academic/technical writing pattern: formal declarative sentence introducing a research Fre lệnh domain with statistical background context. The phrase "Some reasons" signals an upcoming list or explanation of why normal lors hyper parameters are used or S השאלה are used, likely enumerating justifications...	false	false	true	0.605578	0.099891	0.627796	0.840307	0.087998	false	false
arxiv	arxiv/0	23	369	for	false	false	0.958241	271.911438	70.21965	115	<explanation> Academic statistical textbook/review tone establishing foundational concepts before elaborating on justifications or applications of semiparametric models. The text introduces "semiparametric models" and begins listing motivations/explanations, signaling a forthcoming enumeration of reasons for their pop...	false	false	true	2.119709	0.575179	1.192952	1.463722	0.218435	true	true
arxiv	arxiv/0	24	279	the	false	false	0.87543	258.11972	70.97184	130	<explanation> Academic/technical exposition introducing statistical modeling concepts, maintaining formal register with phrases like "it is a generalization" establishing seminar or tutorial context. The text has established an introduction pattern: lasso regression and lasso variables introduced, now logically requir...	false	false	true	4.684563	7.53756	4.750043	0.650675	0.268166	false	false
arxiv	arxiv/0	25	2,393	success	false	true	0.956534	290.499756	70.528061	125	<explanation> Academic/statistical science writing introducing Generalized Generalized Linear Models (GLMs), explaining their Twin Gaussians roots but pivot focus changes. The text poses a question ("why...are important?") and signals an enumerated answer: "basic reasons for the success" strongly implies a list or ser...	false	false	true	0.076003	0.002441	0.062358	0.211532	0.003441	false	false
arxiv	arxiv/0	26	315	of	false	false	0.892911	256.077789	70.426712	133	<explanation> Academic statistical methodology paper, establishing蚂蚁/connections analogy tolinear regression, requiring continuation of formal definition or explanation. The structural metaphor framing expects technical elaboration. The sentence beginning "a major advantage of" sets up an enumerated or explained attri...	false	false	true	0.331616	1.018106	0.64853	0.704787	0.369256	true	false
arxiv	arxiv/0	27	63,238	additive	false	false	0.915629	249.002151	69.102211	104	<explanation> Rhetorical question-answer pattern: text poses question ("Why?") then begins answering it with " ApiResponse" followed by justification. Statistical econometric domain: technical writing about modeling methods, formal academic/introductory tone established by textbook chapter context. Final token " addi...	false	false	true	0.009691	0.000826	0.008217	0.03824	0.011049	false	false
arxiv	arxiv/0	28	4,119	models	false	false	0.905677	258.979218	69.736305	124	<explanation> Academic/technical textbook exposition on statistical methods, establishing additive regression as a topic before pivoting to enumerate advantages. The phrase "main advantages of additive models" signals an upcoming list or enumeration of technical benefits, likely covering overfitting control, flexibili...	false	false	true	0.390188	0.019479	0.272082	0.913982	0.055151	false	false
arxiv	arxiv/0	29	525	are	false	false	0.873551	273.762207	70.488686	97	<explanation> Academic technical exposition establishing a topic/list structure, requiring enumeration of advantages oflinear models in machine learning. The phrase "advantages...are" creates a strong semantic promise of a list of specific benefits or properties, such as simplicity, interpretability, or computational ...	false	false	true	1.867667	3.21841	1.333041	1.677473	0.593399	true	true
arxiv	arxiv/0	30	862	their	false	false	0.934877	275.723053	69.925262	107	<explanation> Academic paper abstract/introduction establishing a topic, requiring justification and elaboration of why modelling trees are relevant. The sentence structure "a key reason is their" promises a specific property or advantage of透视 trees, likely improved interpretability, flexibility, or SP nature compared...	false	false	true	5.524507	10.162697	6.309626	1.858669	0.402524	true	true
arxiv	arxiv/0	31	7,172	increased	false	false	0.945502	264.980713	69.941811	119	<explanation> Academic paper introducing/comparing statistical modeling approaches, establishing formal tone with technical vocabulary like "variable Milito" and " estimation." The sentence structure promises a list of advantages of lasso-based regression; "important advantages are their increased" signals a noun phra...	false	false	true	0.190314	0.079038	0.048874	0.780647	0.35579	false	false
arxiv	arxiv/0	32	24,177	flexibility	false	false	0.934655	283.997498	70.142418	121	<explanation> Academic paper abstract/introduction following standard economics/statistics format, establishing methodology and motivation. Semantic expectation: the phrase "primary appeal is their increased flexibility" signals a list or elaboration of advantages of semiparametric models is coming — likely "and adapt...	false	false	true	6.42367	6.587628	6.041554	0.715449	0.378933	true	true
arxiv	arxiv/0	33	979	when	false	false	0.874629	251.353683	69.962791	87	<explanation> Academic statistical comparison paper establishing a structural claim, expecting elaboration of麻醉 regression's advantages over traditional regression models. The sentence structure "one of the advantages... is" requires a completing predicate describing the mathematical/statistical property and justifica...	false	false	true	0.070096	1.465929	3.259504	1.331043	2.910236	true	true
arxiv	arxiv/0	34	7,707	compared	false	false	0.911039	269.049866	70.583183	117	<explanation> Academic review paper on kernel methods, establishing a balanced pros-and-cons structure with epidemiological applications as context ("controlling variables in complex relationships"). The comparative claim " dictionaries"-phrase signals an incomplete comparison requiring a contrast target, almost certa...	false	false	true	0.253034	0.145788	0.203607	0.0325	0.00688	false	false
arxiv	arxiv/0	35	311	to	false	false	0.933015	268.751984	69.693199	105	<explanation> Academic statistics textbook expository writing establishing comparative analysis of prediction methods, requiring formal technical elaboration. A contrast structure is underway: "mostly because of their flexibility," signaling that trade-offs or advantages will be enumerated (flexibility vs. linear mode...	false	false	true	1.148691	2.087561	0.847335	1.273889	0.299344	true	false
arxiv	arxiv/0	36	13,482	linear	false	false	0.932908	257.105621	68.99292	93	<explanation> Academic course content overview establishing technical/statistical topic-specific register throughout. Enumeration of topics signaled by semicolon introduces a list of course content descriptors for model estimation, implying more topics follow. " Comparing to linear" ends mid-sentence, requiring gramm...	false	false	true	5.215922	6.429081	6.754395	0.060421	0.267061	false	false
arxiv	arxiv/0	37	476	or	false	false	0.951311	255.966965	69.32103	128	<explanation> Academic/statistical paper abstract introducing a new method, requiring continuation of the technical argument and description of limitations of traditional models. The phrase "largerceptive families of dependent variables" signals a contrast structure: flexible multivariate/regression alternatives over ...	false	false	true	1.899221	6.045886	1.213044	4.290754	0.52508	true	false
arxiv	arxiv/0	38	66,117	generalized	false	false	0.945863	252.223282	69.658241	110	<explanation> Academic/statistical software documentation genre, describing a package's capabilities and theoretical basis with formal technical register. The text is building a contrast between overwhelming standard alternatives ("linear or generalized") and implicit superiority of the lasso-based alternative, settin...	false	false	true	0.004996	1.354741	0.007663	1.334932	0.003727	true	false
arxiv	arxiv/0	39	13,482	linear	false	false	0.921777	248.632034	69.677841	122	<explanation> Academic paper abstract introducing a statistical method, with formal technical register and domain-specific terminology like "generalized linear models" and "trees." The abstract is mid-sentence listing methodological advantages of the proposed approach, having already named the method and begun a compa...	false	false	true	0.008866	0.000268	0.02049	0.033661	0.00727	false	false
arxiv	arxiv/0	40	4,119	models	false	false	0.898732	258.867188	70.435028	106	<explanation> Academic abstract/methodology paper tone established, with technical vocabulary like "generalized linear models" and "penalty methods." Parallel comparison structure begun:lassoAdvantages are being enumerated, suggesting additional comparative claims about red家乡 flexibility versus regression/classical mo...	false	false	true	1.31793	1.842369	1.337599	1.892313	1.854469	true	true
arxiv	arxiv/0	41	323	and	false	false	0.92695	255.634155	71.214737	116	<explanation> Academic paper abstract continuing a list of benefits of lasso regression, following "number one is their freedom from curse of dimensionality" structure — more advantages expected next. The text promises enumeration of pros of lasso regression, with "freedom from curse of dimensionality" serving as the ...	false	false	true	0.354443	0.247131	0.452891	0.2149	0.114119	false	false
arxiv	arxiv/0	42	862	their	false	false	0.912681	263.089111	69.837608	123	<explanation> Academic paper introduction establishing lasso's virtues, requiring continuation of listed advantages (overfitting cautiousest, Newtonian optimization nonparametric). The phrase "also thanks to their..." signals an additive list structure enumerating advantages of forests, so a parallel noun phrase descr...	false	false	true	4.472805	2.453857	5.936326	1.306473	0.62564	false	false
arxiv	arxiv/0	43	7,172	increased	false	false	0.933378	263.091217	70.13456	120	<explanation> Academic/statistical textbook register, explaining ensemble methods with formal terminology like "carbon neural networks" and "high-dimensional parameter." Comparative argument structure: lasso i grab neuro UB models are being contrasted with regression, with pros listed ("flexibility," "simpler assumpti...	false	false	true	0.043431	2.480098	1.240703	2.346601	1.084973	true	false
arxiv	arxiv/0	44	14,198	interpret	false	false	0.933437	253.6978	69.629448	123	<explanation> Academic abstract structure (numbered 1, 2) listing advantages of quantile regression; technical mathematical/econometric domain signals continuation of formal prose. The sentence contrasts quantile regression against ordinary least squares using "such" to introduce a consequence clause about interpretab...	false	false	true	0.002136	0.01151	0.001411	0.012635	0.001679	false	false
arxiv	arxiv/0	45	2,897	ability	false	false	0.908683	286.346161	70.41642	115	<explanation> Academic/technical exposition style: written prediction science textbook introducing lasso regression concept systematically and formally. The sentence establishes a continuing list of advantages: "high dimensionality" benefit already stated, "improvable interpretability" completing a paired comparison, ...	false	false	true	1.377821	1.476328	3.220718	0.368195	0.735718	false	false
arxiv	arxiv/0	46	979	when	false	false	0.91491	262.647552	70.458969	100	<explanation> Academic abstract/paper describing ensemble model properties, expecting continuation of comparative advantages over regression orsigmoid models. A list of comparative qualities ("better predictive performance when compared to linear regression models") establishes a rhythmic parallel structure pattern, p...	false	false	true	0.003356	0.000876	0.010296	0.004938	0.00483	false	false
arxiv	arxiv/0	47	7,707	compared	false	false	0.90111	266.291748	70.744675	117	<explanation> Academic statistics textbook register, comparing linear regression's properties against alternatives likeinverse modeling or nonparametric methods. Parallel comparison structure established: "compared to classical linear models" already stated, now a second contrast requires its counterpart (e.g., "compa...	false	false	true	0.01297	0.043954	0.101049	0.015759	0.061745	false	false
arxiv	arxiv/0	48	311	to	false	false	0.902035	261.451263	70.065521	102	<explanation> Academic abstract presenting a method, establishing contrast between models with bias flexibility vs. alternatives like logistic regression or neural networks. The comparison structure requires completing the contrast: "compared to [parameter space model / model selection approach]," a standard machine l...	false	false	true	0.236065	2.553409	1.574378	2.208367	1.028744	true	false
arxiv	arxiv/0	49	7,225	fully	false	false	0.948327	273.67218	69.64518	132	<explanation> Academic statistical paper introducing a method, establishing a contrast between linear and nonlinear regression approaches throughout. The sentence structure contrasts "linear models" versus multiple alternative approaches; "more flexible,... than fully" signals requiring a completion of the named alter...	false	false	true	0.016679	0.967008	0.042819	0.921209	0.027706	true	false
arxiv	arxiv/0	50	2,477	non	false	false	0.942391	232.80394	69.942383	120	<explanation> Statistical/technical academic prose describing a modeling method, requiring continuation of formal terminology about semi-parametric approaches and nonparametric estimation. The sentence establishes a contrast structure ("more flexible models than fully non-"), with the argument building toward justifyi...	false	false	true	0.026522	0.001972	0.015388	0.09256	0.014084	false	false
arxiv	arxiv/0	51	903	param	false	false	0.920786	218.030884	69.733727	109	<explanation> Academic technical documentation tone describing statistical methodology, with structured enumeration of methods and properties. Formal register maintained throughout. The text is mid-explanation of whylassical's superiority over parametric methods, specifically detailing its bias properties ("nonparamet...	false	false	true	0.000408	0.000005	0.000124	0.002576	0.000308	false	false
arxiv	arxiv/0	52	16,340	etric	false	false	0.901401	256.335419	69.350929	122	<explanation> Academic statistics textbook style, describing a classification of statistical methods with formal tone ("most general," "limitations"). The sentence is mid-argument, contrasting model-based methods favorably against nonparametric alternatives, building toward elaboration of specific trade-offs (small ga...	false	false	true	0.072501	0.065288	0.32163	0.214554	0.168074	false	false
arxiv	arxiv/0	53	4,119	models	false	false	0.913325	247.535202	70.870865	121	<explanation> Academic statistics textbook/introduction register, defining a technical learning concept "p models" with formal definitional structure. Comparative framing grammatically and semantically demands completion — "much simpler than predicts in forecasts" parallels the "flexibility compared to fully又一次 explan...	false	false	true	0.697124	3.024722	6.618645	1.481559	3.389766	true	true
arxiv	arxiv/0	54	659	.	false	false	0.81196	244.175156	72.152924	109	<explanation> Academic/statistical methodology writing establishing advantages of an estimation approach, expecting detailed elaboration or tor nàomodels argument. The introduction has introduced article ("article") and key concept, momentum toward explaining limitations, applications on stamps/daming, or formal expos...	false	false	true	4.979106	12.542599	14.131894	6.742978	8.412248	true	true
arxiv	arxiv/0	55	432	it	false	false	0.900547	216.046997	71.519394	118	<explanation> Academic prose explanation of a statistical model (extending lasso in high-dimensional settings), maintaining formal technical register throughout. The second paragraph builds a descriptive overview of lasso's capabilities; the structure follows an introduction pattern explaining mechanism, benefits, and...	false	false	true	0.146684	0.114751	0.28608	0.163873	0.094681	false	false
arxiv	arxiv/0	56	374	is	false	false	0.737287	257.577148	71.717758	124	<explanation> Academic technical writing about statistical regression adjustments, establishing a positive framing with "They are powerful and flexible" suggesting continued elaboration on properties or applications. The passage introduces lasso/polymer as a topic, following an introductory framing typical of Evans an...	false	false	true	1.842821	10.806581	13.854457	7.753957	10.925465	true	true
arxiv	arxiv/0	57	1,632	well	false	false	0.952766	239.187317	70.534431	118	<explanation> Academic or technical exposition on regression/supervised learning, establishing background before discussing limitations or advantages. The rhetorical structure sets up a contrast: having described the "penalty approach" and its applications, the text now pivots to foundational theory, implying "known" ...	false	false	true	5.329769	7.558417	9.3332	0.086395	0.078318	false	false
arxiv	arxiv/0	58	481	-	false	false	0.759899	202.972565	70.704857	124	<explanation> Formal scientific/technical register with academic hedging language ("in many cases", "this transition") constraining precise, measured continuation. The sentence structure "this type of" is incomplete, strongly requiring a noun phrase describing the category of transition behavior (e.g., "types of adapt...	false	false	true	0.022449	0.591952	0.517316	0.615161	0.467017	false	false
arxiv	arxiv/0	59	3,881	known	false	false	0.935589	261.885254	70.145859	110	<explanation> Academic/statistical writing establishing a gap-filling methodology, typical of a paper introduction or section header, predicting technical exposition. The phrase "it is well-known" signals an apparent imminent citation of established results or a fixed principle about thelasso selection criterion, whic...	false	false	true	0.014506	0.055916	0.062017	0.040712	0.035976	false	false
arxiv	arxiv/0	60	429	that	false	false	0.946912	251.590729	70.049629	100	<explanation> Academic statistics paper tone requiring formal, technical mathematical language throughout. The paper contrasts traditional linear models with additive regression models, momentum shifts to supporting theory/results — a claim or bound regarding lasso versatility is expected next. Final token "that" int...	false	false	true	8.291589	11.422976	8.269343	0.489173	0.171649	true	false
arxiv	arxiv/0	61	1,661	good	false	false	0.936412	268.222076	69.835594	118	<explanation> Academic abstract/introduction structure: formal statistical methodology paper establishing leaderboard context, now elaborating on theoretical grounding. Narrative momentum toward justification: "It is known that good" strongly signals a well-known benchmark claim about\Authentic Res hull (e.g., adaptiv...	false	false	true	4.344236	9.357332	3.877472	5.53068	0.627971	true	false
arxiv	arxiv/0	62	14,932	estim	false	false	0.8762	234.110931	69.094177	119	<explanation> Statistical textbook/walkthrough genre with structured academic explanation maintaining formal register throughout. The phrase "proper estim-" explicitly begins a word, almost certainly "estimation," continuing the technical explanation ofחשש regression methodology, likely followed by details on estimati...	false	false	true	0.002161	0.908638	0.023035	0.91183	0.017949	true	false
arxiv	arxiv/0	63	2,973	ators	false	false	0.931669	277.458496	69.685547	108	<explanation> Academic/statistical textbook register with citation formatting (NVPC) signals technical statistical content about high-dimensional regression methods. Narrative momentum: the text introduces a key consequence of high-dimensional regression (" hose restrictions on model dimension"), now building toward d...	false	false	true	3.790416	7.200409	3.857401	1.176598	0.290308	true	true
arxiv	arxiv/0	64	304	in	false	false	0.893749	252.226242	69.59433	129	<explanation> Academic statistics paper abstract, establishing context for a contrast between simple and sophisticated estimation techniques for dependent variables in binstructional data. The argument follows a "complicates approach, but modern methods handle it" structure, where goodness-of-fitting limitations frami...	false	false	true	0.486468	1.452363	2.360379	1.142188	1.023648	true	true
arxiv	arxiv/0	65	63,238	additive	false	false	0.924016	243.103745	69.596596	103	<explanation> Academic technical writing about additive modeling, maintaining formal register and explanatory tone throughout. Narrative momentum: the sentence introduces an incomplete contrast ("Although additive models have some numerical instability..."), signaling an expected counterargument explaining regularizat...	false	false	true	0.004324	0.001698	0.019358	0.01031	0.010748	false	false
arxiv	arxiv/0	66	4,119	models	false	false	0.871238	260.946167	69.462952	119	<explanation> Academic statistical genetics paper establishing theoremidia regression models context, expecting formal methodology or conditioning discussion to follow. The text follows a conditional "if we assume... then" mathematical framing pattern, suggesting a results theorem or identification criterion will comp...	false	false	true	1.608565	0.766202	1.213516	1.06036	0.56675	true	true
arxiv	arxiv/0	67	525	are	false	false	0.864093	266.230865	69.490631	125	<explanation> Academic abstract style: formal, technical tone describing a statistical methodology paper on high-dimensional high-dimensional models. Grammatical/syntactic completion: "are" ends an incomplete predicate ("good estimators for the causal density functions are"), requiring an adjective or noun phrase desc...	false	false	true	6.11639	4.681731	4.981789	2.840975	0.919915	true	true
arxiv	arxiv/0	68	304	in	false	false	0.893771	243.505844	70.315842	100	<explanation> Academic abstract structure: formal statistical methodology paper, establishing theoretical framework before stating results/findings. Causal regression prediction in high-dimensional lasso: the argument builds toward explaining why lasso lasso is criticized or needed, requiring technical justification n...	false	false	true	0.285899	5.934834	0.400178	4.91452	0.123755	true	false
arxiv	arxiv/0	69	4,586	general	false	false	0.864977	276.142487	69.743073	120	<explanation> Academic statistical paper establishing technical scope, tone is formal and specialized — references to semiparametric models and predictors expected to continue. The sentence "Under certain assumptions the optima are in general" promises a theoretical result or claim about notation nonparametric estimat...	false	false	true	4.673921	11.92198	5.749063	5.786506	0.693079	true	false
arxiv	arxiv/0	70	2,686	less	false	false	0.913238	255.41188	70.175018	116	<explanation> Academic statistical writing comparing density estimation frameworks, requiring technical claims about methods like logit, logistic, and constrained fits. The sentence structure "theoretical limit version procedures...are less [adjective]" sets up a contrast or trade-off, likely "less sensitive to" or "l...	false	false	true	4.303673	9.675227	3.477003	7.657244	0.638877	true	false
arxiv	arxiv/0	71	36,997	prone	false	false	0.946145	266.10318	70.461044	118	<explanation> Academic technical writing about sparse linear regression, establishing theoretical/mathematical claims about the L1-f NOTE least squares approach. Argumentative momentum: the sentence has begun listing properties of L1 bacteriological inference; continuation of positive theoretical claims is expected, l...	false	false	true	0.001572	0.001627	0.001747	0.001076	0.000286	false	false
arxiv	arxiv/0	72	311	to	false	false	0.920474	254.830856	69.02771	112	<explanation> Academic statistical textbook, explaining a trade-off about estimation limitation vs. flexibility in nonparametric methods like ML późatem. Comparative property being described: localized النوع Bakan Polırl baskıwró's advantage over global regression, establishing a contrast with historical cubic/quadrat...	false	false	true	0.084687	4.033794	3.429239	3.64977	3.026597	true	true
arxiv	arxiv/0	73	279	the	false	false	0.92167	245.792282	70.176079	122	<explanation> Technical statistical methodology paper, following established pattern of describing regression approaches for spatial prediction modelling. Contrastive comparison momentum: regression using quadratic smooth function has been introduced and its advantages are being enumerated; the phrase sets up a contra...	false	false	true	0.133484	1.837012	0.903756	1.743018	0.634429	true	false
arxiv	arxiv/0	74	40,000	curse	false	false	0.946186	259.876526	69.883018	120	<explanation> Academic paper abstract continuing mid-sentence, establishing methodological contributions with formal register ("we propose," "are proven"). Historical jitter SW analysis pattern: the abstract lists sequential adaptations/improvements ("beautiful," "further improvements"), implying a conclusion about th...	false	false	true	0.000699	0.000703	0.002623	0.003354	0.005081	false	false
arxiv	arxiv/0	75	315	of	false	false	0.866459	249.00058	69.695328	103	<explanation> Academic machine learning paper comparing algorithms, requiring continuation of technical evaluation claims about decision trees vs. logistic regression. The sentence structure promises a contrasting conclusion: sacrifice the curse of [complexity/sparicity is avoided] but also retain accuracy or speed ad...	false	false	true	12.251055	16.893951	8.876244	10.387232	0.001656	true	false
arxiv	arxiv/0	76	1,550	high	false	false	0.944069	240.004105	69.630074	112	<explanation> Academic abstract structure: statistical methodology paper describing a new classification algorithm, establishing formal tone with domain-specific terminology. Technical content momentum: the abstract contrasts "linear discrimination functions" vs "whole functions," building toward resolution of curse-o...	false	false	true	0.282416	0.480887	0.11928	1.562211	1.705986	false	false
arxiv	arxiv/0	77	12,871	dimension	false	false	0.911998	237.945511	69.607162	108	<explanation> Technical academic abstract describing a statistical methodology, formal register with structured description of an estimation approach. The sentence structure promises completion of the explanation of Lasso's capabilities, specifically detailing whyimensionality" problem is resolved despite high-dimensi...	false	false	true	0.000691	0.000271	0.000357	0.003583	0.001664	false	false
arxiv	arxiv/0	78	2,719	ality	false	false	0.897658	261.925262	70.367249	114	<explanation> Academic computer science paper explainingSNs, establishing a comparative argument that contrasts dense vs. sparse models on learning performance. The text builds a contrast structure: dense models suffer from curse of dimensionality, sparse models thrive, implying an elaboration on why sparseness matter...	false	false	true	0.367287	4.633703	3.644085	3.666706	2.814749	true	true
arxiv	arxiv/0	79	1,091	than	false	false	0.868554	249.884064	69.789703	107	<explanation> Academic technical writing on high-dimensional statistical methods, establishing a comparative claim about logistic regression and classification methods. The argument's logical momentum: a statistical efficiency advantage has been introduced, requiring completion with the specific alternative being comp...	false	false	true	2.734709	11.508996	10.353416	2.056377	2.058097	false	true
arxiv	arxiv/0	80	1,661	good	false	false	0.857216	252.040085	70.11013	122	<explanation> Academic statistics textbook register, maintaining formal technical exposition about parameter space classification theory. Narrative/argumentative momentum: the sentence is building a trade-off comparison between high-dimensional models (Lear inequality) and model selective estimation, establishing quan...	false	false	true	0.004584	0.045126	0.169483	0.046127	0.15828	false	false
arxiv	arxiv/0	81	14,932	estim	false	false	0.680386	245.748596	69.430183	126	<explanation> Academic/statistical software documentation describing a method's theoretical Replies pattern, maintaining formal technical prose throughout. The text is systematically explaining the additive model's theoretical foundations, establishing notation before describing algorithms and methods, implying藐/estim...	false	false	true	0.000014	0.000007	0.000326	0.000105	0.000306	false	false
arxiv	arxiv/0	82	2,973	ators	false	false	0.872853	257.205841	69.818932	117	<explanation> Statistical review paper establishing theoretical results, citing high-dimensional models; discussion of methods and performance criteria drives expectation of comparative conclusions. Asymptotic limit behavior contrast: linear models stitched at boundaries are being compared versus general blanket regre...	false	false	true	0.080608	0.442614	1.142531	0.218523	0.829301	false	true
arxiv	arxiv/0	83	304	in	false	false	0.894388	248.10202	69.587547	121	<explanation> Academic abstract in statistics/p econometrics, maintaining formal technical register with precise terminology (exakt, additiv说明, nonparametric modeling). Contrastive construction "compared to the upper=''): 第三者 comparison structure signals completing the contrast between lasso's downward bias and superi...	false	false	true	0.713444	0.526893	1.462186	3.072391	1.614358	false	true
arxiv	arxiv/0	84	7,225	fully	false	false	0.937018	263.850403	70.401634	130	<explanation> Academic statistical methods paper establishing theoretical justification for lasso regression with high-dimensional variables. Technical argument demands conclusion validating lasso's properties. The sentence contrasts lasso with full nonparametric models (pris incluso), following a pattern of paired co...	false	false	true	0.006437	0.019737	0.014079	0.065347	0.036461	false	false
arxiv	arxiv/0	85	2,477	non	false	false	0.887464	214.393158	70.291153	114	<explanation> Technical ML/statistics vocabulary pattern: terms like "linear models," "logistic models," "penalty parameter" signal high-dimensional/statistical methodology discourse. Repetition/continuation: the text is mid-sentence restart, suggesting a revised or duplicated passage recycling prior content about hig...	false	false	true	0.000397	0.000466	0.002695	0.00161	0.002708	false	false
arxiv	arxiv/0	86	903	param	false	false	0.750302	221.570038	69.456589	121	<explanation> Academic abstract structure: thesis stated, methodology described, now expects elaboration of comparative results/advantages of lasso-based models. Domain terminology momentum: machine learning/statistical methods paper on sparse modeling and lasso, expecting technical continuation with specific model na...	false	false	true	0	0.000001	0.000076	0.000001	0.000075	false	false
arxiv	arxiv/0	87	16,340	etric	false	false	0.89331	251.031448	69.447372	112	<explanation> Academic paper abstract establishing a technical argument about regression models-over-modelingCoverage vs. conventional nonparametric alternatives. The argument structure requires completing the contrast: parametric versions outperform nonparametric models, so a conclusion or elaboration on this efficie...	false	false	true	0.048247	0.17153	0.128226	0.349963	0.158667	false	false
arxiv	arxiv/0	88	4,119	models	false	false	0.830323	237.503738	71.433876	119	<explanation> Academic/pedagogical tone explaining econometric vs. machine learning methodology, distinguishing theoretical explanations from practical limitations with formal register throughout. Comparative argument structure:linear models' advantages over nonlinear models are being outlined, with two contrasting be...	false	false	true	0.285592	2.581234	6.660788	2.414687	6.064503	true	true
arxiv	arxiv/0	89	659	.	false	false	0.805975	225.245529	72.481262	119	<explanation> Academic/statistical writing style established, suggesting continuation of technical exposition about linear regression properties or limitations, maintaining formal register. The text has been describing advantages andproperties of spatially dependent regression models, building an argumentative case fo...	false	false	true	7.637235	12.721655	13.668458	6.784811	8.635742	true	true
arxiv	arxiv/0	90	1,657	many	false	false	0.942194	270.040131	70.185806	124	<explanation> Academic/Tex Science Forum post statistical learning theory context, discussing Boost and𝓰 methods for high-dimensional explanatory models. The text is building a structured argument: motivation stated (smoothness of relationship → regression/machine learning), thencore limitation acknowledged. The next...	false	false	true	6.752645	7.176762	5.596021	1.903559	0.544485	true	true
arxiv	arxiv/0	91	10,295	examples	false	false	0.940784	269.316254	70.724419	111	<explanation> Academic statistics paper establishing context for a methodological discussion, expecting evidential or illustrative examples to follow "Many examples." The argument structure:lasso regression concept introduced, claims established (variable selection, p-values, high-dimensionalمزيد), momentum building t...	false	false	true	0.413256	0.425018	0.97669	0.236449	0.269131	false	false
arxiv	arxiv/0	92	315	of	false	false	0.944087	255.420425	69.955536	108	<explanation> Academic statistical course notes establishing ensemblelearning via boosting techniques, following a formal technical exposition pattern with historical framing. Narrative momentum: the text has covered ensemble justification, theory background, and models covered; the next logical move is concrete examp...	false	false	true	2.041919	4.303619	2.798662	0.827859	0.314443	false	false
arxiv	arxiv/0	93	1,741	such	false	false	0.926834	250.251999	69.769165	120	<explanation> Academic abstract structure: introduction of concept/adaptation complete, now moving toward specific examples or results. Formal technical prose maintained throughout. The sentence "A number of instances of such" promises concrete examples of truncated/regular methods like lasso or elastic net, continuin...	false	false	true	1.81697	5.360602	3.260218	0.927657	0.346824	false	false
arxiv	arxiv/0	94	14,932	estim	false	false	0.911591	234.109451	69.460121	119	<explanation> Academic technical writing on high-dimensional practical statistics, maintaining formal register with domain-specific terminology ("censorship," "density estimation," "spiracy"). Narrative momentum: the lecture/expository structure introduces limitations and solutions, now building toward a survey of spe...	false	false	true	0.000061	0.001629	0.000178	0.001721	0.000073	false	false
arxiv	arxiv/0	95	2,973	ators	false	false	0.782033	265.81958	69.948486	113	<explanation> Academic survey/review paper on high-dimensional statistics, maintaining formal technical register with precise terminology like "p variables" and "high-dimensional." The text has introduced a problem (sp looseness) and mentioned causal inference, creating argumentative momentum toward examples of estima...	false	false	true	9.147048	10.832438	9.164675	0.568956	0.581495	false	true
arxiv	arxiv/0	96	9,173	belong	false	false	0.930027	241.464859	70.534561	113	<explanation> Academic survey/tutorial structure introducing robust optimization then practical applications, suggesting a continued enumeration of methods or models. The sentence about models "with a somewhat robust晚会 time mean" establishes that superiority/superior methods exist, momentumpushing toward completing th...	false	false	true	0.002089	0.003059	0.004077	0.002966	0.001177	false	false
arxiv	arxiv/0	97	311	to	false	false	0.904125	246.615326	69.820145	112	<explanation> Academic/statistical textbook introducing regression estimation in high-dimensional or sparse settings, establishing motivation for recent research. Narrative momentum: the text has enumerated benefits of regression over classical marzo models and stated this field is active, with "numerous methods" sign...	false	false	true	0.042868	0.104065	0.211495	0.090494	0.13922	false	false
arxiv	arxiv/0	98	279	the	false	false	0.932235	251.960892	69.963081	100	<explanation> Academic statistical/statistical methodology text establishing a survey of curve estimation approaches, requiring continuation of technical vocabulary. The sentence structure "绝大部分 curve estimation classes belong to the" sets up a categorical taxonomy or classification system (e.g., "class of interval re...	false	false	true	6.69582	9.129344	5.594884	0.579131	0.168342	false	false
arxiv	arxiv/0	99	3,460	large	false	false	0.904425	250.460358	70.718742	112	<explanation> Statistical methodology paper establishing context; next content should continue surveying model/estimator types in formal academic tone. The sentence structure lists high-dimensional variables and regression models, with nested/ PLUGIN.processor approach being introduced — implying further classificatio...	false	false	true	0.730523	0.397465	0.650177	0.147997	0.0519	false	false

End of preview. Expand in Data Studio

No dataset card yet

Downloads last month: 32