source
stringlengths 36
80
| text
stringlengths 51
500
|
|---|---|
https://en.wikipedia.org/wiki/List_of_IBM_products#215
|
the Bureau of the Census 1890-1940. US GPO.
- ^ a b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae af ag ah ai aj ak al am an ao ap aq ar as at Meacham, Alan D. (1961). Data Processing Equipment Encyclopedia. Gille Associates. Book includes photos of some machines
- ^ a b Scans of plates in L.J. Comrie articles from Monthly Notices of the Royal Astronomical Society 1928 and 1932, here
- ^ a b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae af ag ah ai aj ak al am
|
https://en.wikipedia.org/wiki/List_of_IBM_products#216
|
v w x y z aa ab ac ad ae af ag ah ai aj ak al am an ao ap aq ar as at au av aw ax ay az IBM Sales Manual. IBM. pages dated from 1963 to 1974
- ^ a b c d e f g h i j k Fierheller, George A. (2014). Do not fold, spindle or mutilate: the "hole" story of punched cards (PDF). Stewart Pub. p. 25. ISBN 978-1-894183-86-4. Archived (PDF) from the original on 2022-10-09. An accessible book of recollections (sometimes with errors), with photographs and descriptions of many unit record machines.
- ^ a b c
|
https://en.wikipedia.org/wiki/List_of_IBM_products#217
|
criptions of many unit record machines.
- ^ a b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae af ag ah ai Lars Poulsen collected a list of IBM unit record machine types and names. "It was collected over a period of several years from the alt.folklore.computers USENET group. I started out with the ones I knew, and slowly people contributed more items, until we have what you see. I could not point you to a single—or even a few—lists with attributions; it was a community effort." –
|
https://en.wikipedia.org/wiki/List_of_IBM_products#218
|
s with attributions; it was a community effort." – Lars Poulsen That list is here
- ^ a b "IBM 29 card punch". IBM Archives. IBM. 2003-01-23. Archived from the original on September 9, 2006.
- ^ "IBM 31 Alphabetical Duplicating Punch". IBM Archives. IBM. 2003-01-23. Archived from the original on January 24, 2005.
- ^ "IBM Type 032 Alphabetical Printing Punch". IBM Archives. IBM. 2003-01-23. Archived from the original on March 21, 2005.
- ^ a b c d e f g h Linnekin, Leroy Corliss (1938). The Scop
|
https://en.wikipedia.org/wiki/List_of_IBM_products#219
|
d e f g h Linnekin, Leroy Corliss (1938). The Scope of Punched Card Accounting. Boston University, College of Business Administration - Thesis.
- ^ a b c d e Machine Methods of Accounting, IBM, 1936
- ^ "IBM 40 tape controlled card punch". IBM Archives. IBM. 2003-01-23. Archived from the original on January 1, 2007.
- ^ a b c d e f g h i j k l m n o p q r s t u v w x y z "IBM Card Equipment Summary" (PDF). 1957.
- ^ "IBM 797 Document Numbering Punch". IBM Archives. IBM. 2003-01-23. Archived from
|
https://en.wikipedia.org/wiki/List_of_IBM_products#220
|
nch". IBM Archives. IBM. 2003-01-23. Archived from the original on December 30, 2006.
- ^ Machine Functions (PDF). IBM. 1957. 224-8208-3. Archived (PDF) from the original on 2022-10-09.
- ^ "Port-A-Punch". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "Votomatic". IBM Archive. IBM. 2003-01-23. Archived from the original on February 26, 2006.
- ^ "Votamatic". Verified Voting Foundation. Archived from the original on 7 August 2020. Retrieved 30 May 2015.
- ^ "I
|
https://en.wikipedia.org/wiki/List_of_IBM_products#221
|
al on 7 August 2020. Retrieved 30 May 2015.
- ^ "IBM Archives: Artifacts list for vol. 2". www.ibm.com. January 23, 2003.
- ^ "Type 070 vertical sorters". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ "(IBM) Hollerith 070 Vertical Sorter". IBM Archives. IBM. 2003-01-23. Archived from the original on January 14, 2005.
- ^ "IBM Type 71 Vertical Sorter". IBM Archives. IBM. 2003-01-23. Archived from the original on January 14, 2005.
- ^ "IBM Type 80 Electric Punc
|
https://en.wikipedia.org/wiki/List_of_IBM_products#222
|
n January 14, 2005.
- ^ "IBM Type 80 Electric Punched Card Sorting Machine". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ a b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae af ag ah IBM Sales Manual, DP Machines, page 1.20, May 1979
- ^ a b IBM 101 Electronic Statistical Machine, A22-0502-0
- ^ Taube, Mortimer (1962). "M. Taube: Experiments with the IBM-9900 and a discussion of an improved Comac as suggested by these experiments". Journal of
|
https://en.wikipedia.org/wiki/List_of_IBM_products#223
|
ac as suggested by these experiments". Journal of Chemical Documentation. 2: 22–26. doi:10.1021/c160004a007.
- ^ Jerry Norman. "The Coordinate and Uniterm Indexing Systems".
- ^ "IBM 77 electric punched card collator". IBM Archives. IBM. 2003-01-23. Archived from the original on September 9, 2006.
- ^ "IBM 85 collator". IBM Archives. IBM. 2003-01-23. Archived from the original on February 7, 2008.
- ^ "IBM 087". IBM Archives. IBM. 2003-01-23. Archived from the original on October 11, 2013.
- ^ "
|
https://en.wikipedia.org/wiki/List_of_IBM_products#224
|
hived from the original on October 11, 2013.
- ^ "IBM 514". IBM Archives. IBM. 2003-01-23. Archived from the original on March 10, 2007.
- ^ "IBM 523 gang summary punch". IBM Archives. IBM. 2003-01-23. Archived from the original on February 7, 2008.
- ^ a b c d e f g h i j k "IBM Operators' Guide" (PDF). IBM. A24-1010-0.
- ^ "IBM 550". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "Hollerith tabulator and sorter". IBM Archives. IBM. 2003-01-23. Archived from
|
https://en.wikipedia.org/wiki/List_of_IBM_products#225
|
er". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ "IBM Archives: Artifacts list for vol. 1". www.ibm.com. January 23, 2003.
- ^ a b c d e f g h i j k l m n o p q r s "IBM Tabulators and Accounting Machines". www.columbia.edu.
- ^ "IBM 285 electric accounting machine". IBM Archives. IBM. 2003-01-23. Archived from the original on January 1, 2007.
- ^ "IBM 301 Accounting Machine". IBM Archives. IBM. 2003-01-23. Archived from the original on August 26, 2006.
- ^
|
https://en.wikipedia.org/wiki/List_of_IBM_products#226
|
Archived from the original on August 26, 2006.
- ^ Pugh, Emerson W. (1995). Building IBM: Shaping and Industry and Its Technology. MIT Press. p. 50. ISBN 978-0-262-16147-3.
- ^ Pugh (1995) pp.50–51
- ^ "The Naval Observatory Table Printer 1945". www.columbia.edu.
- ^ a b c d e f IBM 402, 403 and 407; IBM 421, 444, 407 and 447 for World Trade Corporation; Computing Accounting Machines. IBM. A24-3475-0.
- ^ IBM 402, 403 and 419 Accounting Machine. IBM. 22-5654-12.
- ^ a b IBM Sales Manual, 11-10-5
|
https://en.wikipedia.org/wiki/List_of_IBM_products#227
|
IBM. 22-5654-12.
- ^ a b IBM Sales Manual, 11-10-55
- ^ Catalog of Copyright Entries. Third Series: 1962: July–December By Library of Congress. Copyright Office, p.1517
- ^ "IBM 405 electric punched card accounting machine". IBM Archives. IBM. 2003-01-23. Archived from the original on September 11, 2006.
- ^ a b c "IBM Archives: Artifacts list for vol. 3, items I-L". www.ibm.com. January 23, 2003.
- ^ a b IBM Operator's Guide. IBM. 1955. 22-8485-3.
- ^ "IBM 407 accounting machine". IBM Archives.
|
https://en.wikipedia.org/wiki/List_of_IBM_products#228
|
3.
- ^ "IBM 407 accounting machine". IBM Archives. IBM. 2003-01-23. Archived from the original on January 1, 2007.
- ^ IBM Sales Manual, 1-2-57
- ^ "IBM Archives: Endicott chronology - 1951-1959". www.ibm.com. January 23, 2003.
- ^ Based on references such as 407-408-409 in IBM Functional Wiring PRinciples A24-1007-0, the 409 is a version of the 407
- ^ IBM 407 Accounting Machine. IBM. A24-1011-0.
- ^ "IBM Machine Load Computer Slide Rule". National Museum of American History. Retrieved 2024-12-
|
https://en.wikipedia.org/wiki/List_of_IBM_products#229
|
nal Museum of American History. Retrieved 2024-12-02.
- ^ a b Bashe, Charles J.; et al. (1986). IBM's Early Computers. MIT. ISBN 978-0-262-02225-5.
- ^ a b c d e f g h i j k l m n "IBM Calculators". www.columbia.edu.
- ^ a b "IBM Relay Calculator". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM 603 electronic multiplier". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ a b "Card-Programmed Calculator". IBM Archives. IBM.
|
https://en.wikipedia.org/wiki/List_of_IBM_products#230
|
b "Card-Programmed Calculator". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ Research, United States Office of Naval (1953). A survey of automatic digital computers. Office of Naval Research, Dept. of the Navy. p. 48.
- ^ "IBM 608 calculator". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ IBM Sales Manual 1979 p.M 1.30
- ^ a b IBM WTC 212-9924-0
- ^ a b "International Time Recording Co. catalog" (PDF)., 1935 or earlier
-
|
https://en.wikipedia.org/wiki/List_of_IBM_products#231
|
Recording Co. catalog" (PDF)., 1935 or earlier
- ^ "IBM 1956: History of the Time Equipment Division and its Products" (PDF). Archived from the original (PDF) on September 14, 2006.
- ^ "IBM: CONSOLIDATED LISTING OF IBM TIME & WEIGHING EQUIPMENT" (PDF). Archived from the original (PDF) on May 3, 2006.
- ^ "IBM 1958: Press release announcing the sale of the domestic time equipment (clocks et al.) business to Simplex Time Recorder Company" (PDF). Archived from the original (PDF) on September 9, 2
|
https://en.wikipedia.org/wiki/List_of_IBM_products#232
|
Archived from the original (PDF) on September 9, 2006.
- ^ "Remote Control Keyboard". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Model 01 electric typewriter". IBM Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ a b c d e f g h i j k l "IBM Archives: IBM typewriter milestones". www.ibm.com. 2003-01-23. Retrieved 2022-11-20.
- ^ "Model A Standard Electric Typewriter". IBM Archives. IBM. 2003-01-23. Archived from the origi
|
https://en.wikipedia.org/wiki/List_of_IBM_products#233
|
Archives. IBM. 2003-01-23. Archived from the original on January 22, 2005.
- ^ "IBM Selectric Typewriter". IBM Archives. IBM. 2003-01-23. Archived from the original on April 8, 2006.
- ^ a b c d e f g h "IBM Archives: IBM typewriter milestones - page 2". www.ibm.com. 2003-01-23. Retrieved 2022-11-20.
- ^ "IBM Executary Model 212". IBM Archives. IBM. 2003-01-23. Archived from the original on February 8, 2008.
- ^ a b c d "IBM Service List" (PDF). Archived (PDF) from the original on 2022-10-09.
-
|
https://en.wikipedia.org/wiki/List_of_IBM_products#234
|
Archived (PDF) from the original on 2022-10-09.
- ^ "IBM Office Products Division highlights: Page 2". IBM Archives. IBM. 2003-01-23. Archived from the original on July 17, 2012.
- ^ "New I.B.M. Copier". The New York Times. May 14, 1982. Retrieved October 9, 2013.
- ^ a b c "IBM Series III Copier/Duplicator Model 85" (Announcement letter). 1986-06-16.
- ^ "1988". IBM Archives: 1980s. IBM. 2003-01-23. Archived from the original on December 16, 2004.
- ^ a b c d e Pugh, Emerson W. (1995). Building
|
https://en.wikipedia.org/wiki/List_of_IBM_products#235
|
4.
- ^ a b c d e Pugh, Emerson W. (1995). Building IBM: Shaping an Industry and Its Technology. MIT. ISBN 978-0-262-16147-3.
- ^ "IBM 805 Test Scoring Machine". IBM Archives. IBM. 2003-01-23. Archived from the original on January 14, 2005.
- ^ "IBM 1958: Press release announcing the sale of the domestic time equipment (clocks et al.) business to Simplex Time Recorder Company" (PDF) (Press release). IBM. 1958. Archived from the original (PDF) on September 9, 2006.
- ^ Catalog of Copyright Entries
|
https://en.wikipedia.org/wiki/List_of_IBM_products#236
|
eptember 9, 2006.
- ^ Catalog of Copyright Entries. Third Series: 1962: July–December By Library of Congress. Copyright Office, p.1518
- ^ "The IBM Lectern". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Radiotype". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "Scanistor". IBM Archives. IBM. 2003-01-23. Archived from the original on December 29, 2004.
- ^ "IBM Shoebox". IBM Archives. IBM. 2003-01-23. Archived from th
|
https://en.wikipedia.org/wiki/List_of_IBM_products#237
|
". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Ticketograph". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Toll Collection System". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Wireless Translation System". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Instruments Hydrogen Peroxide Analyzer". IBM Archives. IBM. 2003-01-23. Arc
|
https://en.wikipedia.org/wiki/List_of_IBM_products#238
|
xide Analyzer". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Industrial Products PW 200 Percussive Welder". IBM Archives. IBM. 2003-01-23. Archived from the original on January 19, 2005.
- ^ "IBM Industrial Scale". IBM Archives. IBM. 2003-01-23. Archived from the original on January 6, 2005.
- ^ a b "FAQ's for Products and Services". IBM Archives. IBM. Archived from the original on December 17, 2004.
- ^ "Electric scoreboards (1949)". IBM Archives. IBM.
|
https://en.wikipedia.org/wiki/List_of_IBM_products#239
|
"Electric scoreboards (1949)". IBM Archives. IBM. 2003-01-23. Archived from the original on August 20, 2010.
- ^ Cockfield, Bryan (2021-10-13). "IBM Cheese Cutter Restoration". Hackaday. Retrieved 2024-05-08.
- ^ "IBM Cheese Cutter". Keswick Ridge Historical Society. Retrieved 2024-05-08.
- ^ "IBM 305 RAMAC". IBM Archives. IBM. 2003-01-23. Archived from the original on January 26, 2013.
- ^ a b IBM 305 Reference Manual, A26-3502-0, 1958
- ^ "IBM 610 Auto-Point Computer". IBM Archives. IBM. 2003
|
https://en.wikipedia.org/wiki/List_of_IBM_products#240
|
610 Auto-Point Computer". IBM Archives. IBM. 2003-01-23. Archived from the original on September 10, 2006.
- ^ "650 Photo album". IBM Archives. IBM. 2003-01-23. Archived from the original on March 12, 2005.
- ^ "IBM 355 Disk Storage". IBM Archives. IBM. 2003-01-23. Archived from the original on January 14, 2005.
- ^ "IBM 533 Card Read Punch". IBM Archives. IBM. 2003-01-23. Archived from the original on November 20, 2005.
- ^ "IBM 537 Card Read Punch". IBM Archives. IBM. 2003-01-23. Archived fro
|
https://en.wikipedia.org/wiki/List_of_IBM_products#241
|
unch". IBM Archives. IBM. 2003-01-23. Archived from the original on November 30, 2005.
- ^ "IBM 652 Control Unit". IBM Archives. IBM. 2003-01-23. Archived from the original on November 20, 2005.
- ^ "IBM 653 Auxiliary Unit". IBM Archives. IBM. 2003-01-23. Archived from the original on November 29, 2005.
- ^ "IBM 650 Data Processing System Bulletin: Immediate Access Storage, Indexing Register, Automatic Floating-Decimal Arithmetic, and Magnetic Tape" (PDF). IBM. Archived (PDF) from the original o
|
https://en.wikipedia.org/wiki/List_of_IBM_products#242
|
pe" (PDF). IBM. Archived (PDF) from the original on 2022-10-09.
- ^ "IBM 650 MDDPM Additional Features: Indexing Accumulators, Floating-Decimal Arithmetic" (PDF). IBM. Archived (PDF) from the original on 2022-10-09.
- ^ "IBM 655 Power Unit". IBM Archives. IBM. 23 January 2003. Archived from the original on November 20, 2005.
- ^ "IBM 650 installation With IBM 727 Magnetic Tape Unit and IBM 355 Disk Storage". IBM Archives. IBM. 23 January 2003. Archived from the original on November 21, 2005.
- ^
|
https://en.wikipedia.org/wiki/List_of_IBM_products#243
|
chived from the original on November 21, 2005.
- ^ "IBM 838 Inquiry Station". IBM Archives. IBM. 23 January 2003. Archived from the original on November 22, 2005.
- ^ "IBM 701". IBM Archives. IBM. 23 January 2003. Archived from the original on December 28, 2004.
- ^ "IBM 706 Electrostatic storage unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ a b c "IBM 711 Punched card reader". IBM Archives. IBM. 23 January 2003. Archived from the original on Dece
|
https://en.wikipedia.org/wiki/List_of_IBM_products#244
|
3 January 2003. Archived from the original on December 17, 2004.
- ^ a b c "IBM 716 Printer". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ a b c "IBM 721 Punched card recorder". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "IBM 726 magnetic tape reader/recorder". IBM Archives. IBM. 23 January 2003. Archived from the original on September 10, 2006.
- ^ a b c d e f g h "IBM 727 Magnetic tape unit". IBM Archive. I
|
https://en.wikipedia.org/wiki/List_of_IBM_products#245
|
f g h "IBM 727 Magnetic tape unit". IBM Archive. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "IBM 731 Magnetic drum reader/recorder". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "IBM 736 Power frame #1". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ a b c "IBM 737 Magnetic core storage unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ a
|
https://en.wikipedia.org/wiki/List_of_IBM_products#246
|
ived from the original on January 19, 2005.
- ^ a b c "IBM 740 Cathode ray tube output recorder". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "IBM 741 Power frame #2". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "IBM 746 Power distribution unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "IBM 753 Magnetic tape control unit". IBM Archives. IBM. 23 January 2003. Arc
|
https://en.wikipedia.org/wiki/List_of_IBM_products#247
|
rol unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ "702 Data Processing System". IBM Archives. IBM. 23 January 2003. Archived from the original on April 21, 2005.
- ^ "704 Data Processing System". IBM Archives. IBM. 23 January 2003. Archived from the original on January 14, 2005.
- ^ "705 Data Processing System". IBM Archives. IBM. 23 January 2003. Archived from the original on January 14, 2005.
- ^ "Reference Manual, IBM 705 Data Processing System
|
https://en.wikipedia.org/wiki/List_of_IBM_products#248
|
"Reference Manual, IBM 705 Data Processing System" (PDF). IBM. 1959. Archived (PDF) from the original on 2022-10-09.
- ^ "709 Data Processing System". IBM Archives. IBM. 23 January 2003. Archived from the original on January 14, 2005.
- ^ a b c d e f g h i j k l m n o p q r s t u "IBM 1401 and 1460 Bibliography" (PDF). A24-1495-4. Archived (PDF) from the original on 2022-10-09.
- ^ a b "IBM 1401 System Summary" (PDF). IBM. September 1964. pp. 9–10. A24-1401-1. Archived (PDF) from the original o
|
https://en.wikipedia.org/wiki/List_of_IBM_products#249
|
10. A24-1401-1. Archived (PDF) from the original on 2022-10-09.
- ^ "7090 Data Processing System". IBM Archives. IBM. 23 January 2003. Archived from the original on January 21, 2005.
- ^ a b "IBM 2361 core storage unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "IBM 5103 Dot matrix printer". computermuseum.informatik.uni-stuttgart.de. Retrieved 2023-05-31.
- ^ IBM 5103 printer. 1979.
- ^ "System/34". IBM Archives. IBM. 23 January 2003. Archived from
|
https://en.wikipedia.org/wiki/List_of_IBM_products#250
|
IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ a b c Elliott, Jim (2004-08-17). "The Evolution of IBM Mainframes and VM" (PDF). SHARE Session 9140. Retrieved 2007-10-21. Slide 28: "9672 to zSeries".
- ^ "IBM zEnterprise BC12 (zBC12) extends modern mainframe capabilities for enterprises of all sizes to help cut costs, improve service, and fuel innovation for business growth" (Announcement letter). 23 July 2013. IBM
- ^ "The IBM z13" (Announcement letter).
|
https://en.wikipedia.org/wiki/List_of_IBM_products#251
|
2013. IBM
- ^ "The IBM z13" (Announcement letter). 14 January 2015. IBM
- ^ "IBM PS/1". IBM Archives. IBM. 23 January 2003. Archived from the original on April 1, 2007.
- ^ "ThinkPads". IBM Archives. IBM. 23 January 2003. Archived from the original on January 21, 2005.
- ^ IBM System 360/Model 20 System Summary, p. 13
- ^ A1 500 CPM Read-only, A2 Punch 500 CPM, A3 Punch 300 CPM
- ^ IBM World Trade. "IBM 1017 Paper Tape Reader, 1018 Paper Tape Punch, 2826 Paper Tape control Offer Increased Capabi
|
https://en.wikipedia.org/wiki/List_of_IBM_products#252
|
ch, 2826 Paper Tape control Offer Increased Capabilities" (PDF). bitsavers.org. Retrieved June 16, 2024.
- ^ IBM System/370 System Summary (PDF). IBM. December 1975. p. 7-51. GA22-7001-4. Archived (PDF) from the original on 2022-10-09.
- ^ "IBM Archives: Microfilm products". www.ibm.com. 2003-01-23. Retrieved 2020-07-03.
- ^ "Microfilm Newsletter 09-1969 002". Issuu. 5 August 2019. Retrieved 2020-07-03.
- ^ a b Administrative Management 1969-04: Vol 30 Iss 4. Dalton Communications, Inc. 1969. p.
|
https://en.wikipedia.org/wiki/List_of_IBM_products#253
|
Vol 30 Iss 4. Dalton Communications, Inc. 1969. p. 45 – via Internet Archive.
- ^ IBM Corporation (1961). 1401 DATA PROCESSING SYSTEM BULLETIN : 1404 PRINTER (PDF). Archived (PDF) from the original on 2022-10-09. Retrieved 2020-03-02.
- ^ "IBM 3800 printer". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ R. C. Miller (May 1984). "Introduction to the IBM 3800 Printing Subsystem Models 3 and 8". IBM Journal of Research and Development. 28 (3): 252–256. doi:
|
https://en.wikipedia.org/wiki/List_of_IBM_products#254
|
of Research and Development. 28 (3): 252–256. doi:10.1147/rd.283.0252.
- ^ a b Baker, F. T. (1972). "Chief programmer team management of production programming" (PDF). IBM Systems Journal. 11 (1): 58. doi:10.1147/sj.111.0056. Archived from the original (PDF) on 2005-01-10.
- ^ a b IBM 50 Magnetic Data Inscriber, Component Description. IBM. 1969. A27-2725-2.
- ^ System z End-to-End Extended Distance Guide (PDF). IBM. March 2014. p. 160.
Physical layer switches for computer communication have been
|
https://en.wikipedia.org/wiki/List_of_IBM_products#255
|
ayer switches for computer communication have been used for a long time. For example, the IBM 2914 bus-and-tag crosspoint switch was available in the 1970s.
- ^ IBM 3174 Establishment Controller Introduction (PDF) (First ed.). IBM. April 1989. p. 2-21. GA27-3750-0.
- ^ IBM Booklet 520-1122, titled IBM 1015, 2250, 2260 Displays: inside has 1015 Inquiry Display Terminal, 2260 Display Station, 2250 Display Unit
- ^ "IBM 1015" (PDF). Archived (PDF) from the original on 2022-10-09.
- ^ "Systems Refer
|
https://en.wikipedia.org/wiki/List_of_IBM_products#256
|
rom the original on 2022-10-09.
- ^ "Systems Reference Library IBM System/360" (PDF). Archived (PDF) from the original on 2022-10-09.
may be attached to an IBM 1016 Control Unit or an IBM 1414 Input/Output Synchonizer
- ^ "IBM 2217 Multiprotocol Concentrator Release 3 and Multiport Feature Supports Client/Server Protocols". ibm.com (Announcement letter). IBM. October 8, 1996. Retrieved January 19, 2025.
- ^ "IBM DPD Chronology". IBM. 23 January 2003.
- ^ "IBM Archives: Rochester chronology - pag
|
https://en.wikipedia.org/wiki/List_of_IBM_products#257
|
003.
- ^ "IBM Archives: Rochester chronology - page 2". IBM. 2003-01-23. Retrieved 2023-04-06.
- ^ "7170 Device Attachment Control Unit Reference and Operations Manual" (PDF) (Third ed.). IBM. October 1984. SA24-4025-0.
- ^ "Binary Dinosaurs - IBM 5150 and 5161". www.binarydinosaurs.co.uk.
- ^ "IBM 803 Proof Machine". IBM Archives. IBM. 23 January 2003. Archived from the original on January 22, 2005.
- ^ Introduction to IBM Data Processing Systems. IBM. 1960. p. 28.
- ^ "1980s". IBM Archives. IB
|
https://en.wikipedia.org/wiki/List_of_IBM_products#258
|
s. IBM. 1960. p. 28.
- ^ "1980s". IBM Archives. IBM. 23 January 2003. Archived from the original on December 16, 2004. Retrieved 2013-10-26.
- ^ "IBM Unveils Drafting System". Computerworld: 57. Dec 13, 1982.
- ^ General Information Manual for the IBM Displaywriter System (PDF) (Fourth ed.). IBM. October 1981. G544-0851-3.
- ^ IBM 6361 Mag Card Unit Customer Setup Guide (PDF) (First ed.). IBM. July 1981. G544-2002-0.
- ^ IBM Sales Manual, May 79
- ^ Longo, Alexander A.; Giunti, Frank E. (August
|
https://en.wikipedia.org/wiki/List_of_IBM_products#259
|
- ^ Longo, Alexander A.; Giunti, Frank E. (August 8–10, 1972). A Sequential Evaluation of Computer Assisted Instruction in US Army Basic Electronics Training. Annual Convention of the Association for the Development of Instructional Systems. Cap Rouge, Quebec.
- ^ "The IBM 4020 Military Computer" (PDF). IBM Federal Systems Division. 31 October 1959. Retrieved September 28, 2009.
- ^ "Gemini computer". IBM Archives. IBM. 23 January 2003. Archived from the original on December 17, 2004.
- ^ "Satur
|
https://en.wikipedia.org/wiki/List_of_IBM_products#260
|
from the original on December 17, 2004.
- ^ "Saturn guidance computer". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "Saturn instrument unit". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "IBM and Skylab". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "IBM and the Space Shuttle". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "TR
|
https://en.wikipedia.org/wiki/List_of_IBM_products#261
|
ved from the original on January 19, 2005.
- ^ "TRACTOR". IBM Archives. IBM. 23 January 2003. Archived from the original on January 19, 2005.
- ^ "IBM 1001 DATA TRANSMISSION SYSTEM (Dec, 1961)". Modern Mechanix. Archived from the original on 2020-07-23. Retrieved 2020-07-23.
- ^ "IBM 1710 industrial control system". IBM Archives. IBM. 23 January 2003. Archived from the original on March 11, 2007.
- ^ "IBM 1800 data acquisition and control system". IBM Archives. IBM. 23 January 2003. Archived fro
|
https://en.wikipedia.org/wiki/List_of_IBM_products#262
|
. IBM Archives. IBM. 23 January 2003. Archived from the original on June 5, 2008.
- ^ a b c "IBM Series/1". IBM Archives. IBM. 23 January 2003. Archived from the original on January 3, 2013.
- ^ "IBM Maritime Applikcation/Bridge System" (PDF). Archived (PDF) from the original on 2022-10-09.
- ^ Data Processor 1974-02: Vol 17 Iss 1. International Business Machines Corporation. 1974. p. 5 – via Internet Archive.
- ^ IBM System370 System Summary (PDF). IBM. 1975. pp. 8–42.1. Archived (PDF) from the
|
https://en.wikipedia.org/wiki/List_of_IBM_products#263
|
F). IBM. 1975. pp. 8–42.1. Archived (PDF) from the original on 2022-10-09.
- ^ "1985". IBM Archives: 1980s. IBM. 23 January 2003. Archived from the original on December 16, 2004.
- ^ "The Columbia Difference Tabulator 1931". www.columbia.edu.
- ^ "IBM SSEC". www.columbia.edu.
- ^ "IBM Research: Dave Ferrucci at Computer History Museum: How it all began and what's next". ibmresearchnews.blogspot.com.
- ^ "IBM 3663 supermarket terminal". IBM Archives. IBM. 23 January 2003. Archived from the origin
|
https://en.wikipedia.org/wiki/List_of_IBM_products#264
|
es. IBM. 23 January 2003. Archived from the original on February 5, 2008.
- ^ "IBM Point-of-sale systems and kiosks from Retail Store Solutions - products and services portfolio". IBM. Archived from the original on August 19, 2006.
- ^ "IBM SurePOS 300 Series - Overview". IBM. Archived from the original on August 22, 2006.
- ^ "IBM SurePOS 500 Series - Overview". IBM. Archived from the original on August 22, 2006.
- ^ "IBM SurePOS 700 Series - Overview". IBM. Archived from the original on August
|
https://en.wikipedia.org/wiki/List_of_IBM_products#265
|
erview". IBM. Archived from the original on August 30, 2006.
- ^ "IBM AnyPlace POS solution - Overview". IBM. Archived from the original on January 25, 2013.
- ^ "A History of BART: The Project is Rescued | bart.gov". www.bart.gov.
- ^ "IBM'S TOUCHMOBILE HELPS FIELD WORKERS COLLECT DATA AT THE TOUCH OF A FINGER". The Free Library by Farlex. Archived from the original on 11 December 2017. Retrieved 10 December 2017.
- ^ a b "IBM Announcement of Program Products/SHARE 1969" (PDF). June 23, 1969. A
|
https://en.wikipedia.org/wiki/List_of_IBM_products#266
|
ogram Products/SHARE 1969" (PDF). June 23, 1969. Archived (PDF) from the original on 2022-10-09.
- ^ IBM System/360 Disk Operating System - Vocabulary File Utility Program for the IBM 7772 Audio Response Unit, Program Number 360N-UT-472 (PDF). IBM. February 1968. GC27-6924-2. Archived (PDF) from the original on 2022-10-09.
- ^ "IBM Content Manager OnDemand". IBM.
- ^ "DCF V1R4.0 Documentation Bookshelf", IBM, August 8, 2001[permanent dead link]
- ^ "BookMaster V1R4.0 Bookshelf product on Printin
|
https://en.wikipedia.org/wiki/List_of_IBM_products#267
|
^ "BookMaster V1R4.0 Bookshelf product on Printing and Publishing CD", IBM, August 1, 1995
- ^ "BookManager Bookshelf", IBM, 2005[permanent dead link]
- ^ "i2 is now part of IBM". IBM web site. Archived from the original on October 2, 2013. Retrieved September 29, 2013.
- ^ "Screen Definition Facility II". ibm.com. IBM. Archived from the original on June 9, 2023. Retrieved 25 February 2017.
- ^ "IBM Computer". St. Petersburg Independent. June 11, 1968.
- ^ "IBM buys Silverpop, cloud-based acqui
|
https://en.wikipedia.org/wiki/List_of_IBM_products#268
|
, 1968.
- ^ "IBM buys Silverpop, cloud-based acquisitions pool dries". adweek.com. Retrieved November 21, 2017.
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#0
|
Natural-language user interface
Natural-language user interface (LUI or NLUI) is a type of computer human interface where linguistic phenomena such as verbs, phrases and clauses act as UI controls for creating, selecting and modifying data in software applications.
In interface design, natural-language interfaces are sought after for their speed and ease of use, but most suffer the challenges to understanding wide varieties of ambiguous input.[1] Natural-language interfaces are an active area of
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#1
|
Natural-language interfaces are an active area of study in the field of natural-language processing and computational linguistics. An intuitive general natural-language interface is one of the active goals of the Semantic Web.
Text interfaces are "natural" to varying degrees. Many formal (un-natural) programming languages incorporate idioms of natural human language. Likewise, a traditional keyword search engine could be described as a "shallow" natural-language user interface.
Overview
[edit]A
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#2
|
natural-language user interface.
Overview
[edit]A natural-language search engine would in theory find targeted answers to user questions (as opposed to keyword search). For example, when confronted with a question of the form 'which U.S. state has the highest income tax?', conventional search engines ignore the question and instead search on the keywords 'state', 'income' and 'tax'. Natural-language search, on the other hand, attempts to use natural-language processing to understand the nature
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#3
|
ural-language processing to understand the nature of the question and then to search and return a subset of the web that contains the answer to the question. If it works, results would have a higher relevance than results from a keyword search engine, due to the question being included.[citation needed]
History
[edit]Prototype Nl interfaces had already appeared in the late sixties and early seventies.[2]
- SHRDLU, a natural-language interface that manipulates blocks in a virtual "blocks world"
-
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#4
|
t manipulates blocks in a virtual "blocks world"
- Lunar, a natural-language interface to a database containing chemical analyses of Apollo 11 Moon rocks by William A. Woods.
- Chat-80 transformed English questions into Prolog expressions, which were evaluated against the Prolog database. The code of Chat-80 was circulated widely, and formed the basis of several other experimental Nl interfaces. An online demo is available on the LPA website.[3]
- ELIZA, written at MIT by Joseph Weizenbaum betwe
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#5
|
- ELIZA, written at MIT by Joseph Weizenbaum between 1964 and 1966, mimicked a psychotherapist and was operated by processing users' responses to scripts. Using almost no information about human thought or emotion, the DOCTOR script sometimes provided a startlingly human-like interaction. An online demo is available on the LPA website.[4]
- Janus is also one of the few systems to support temporal questions.
- Intellect from Trinzic (formed by the merger of AICorp and Aion).
- BBN’s Parlance buil
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#6
|
merger of AICorp and Aion).
- BBN’s Parlance built on experience from the development of the Rus and Irus systems.
- IBM Languageaccess
- Q&A from Symantec.
- Datatalker from Natural Language Inc.
- Loqui from BIM Systems.
- English Wizard from Linguistic Technology Corporation.
Challenges
[edit]Natural-language interfaces have in the past led users to anthropomorphize the computer, or at least to attribute more intelligence to machines than is warranted. On the part of the user, this has led t
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#7
|
warranted. On the part of the user, this has led to unrealistic expectations of the capabilities of the system. Such expectations will make it difficult to learn the restrictions of the system if users attribute too much capability to it, and will ultimately lead to disappointment when the system fails to perform as expected as was the case in the AI winter of the 1970s and 80s.
A 1995 paper titled 'Natural Language Interfaces to Databases – An Introduction', describes some challenges:[2]
- Modi
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#8
|
ntroduction', describes some challenges:[2]
- Modifier attachment
- The request "List all employees in the company with a driving licence" is ambiguous unless you know that companies can't have driving licences.
- Conjunction and disjunction
- "List all applicants who live in California and Arizona" is ambiguous unless you know that a person can't live in two places at once.
- Anaphora resolution
- resolve what a user means by 'he', 'she' or 'it', in a self-referential query.
Other goals to cons
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#9
|
, in a self-referential query.
Other goals to consider more generally are the speed and efficiency of the interface, in all algorithms these two points are the main point that will determine if some methods are better than others and therefore have greater success in the market. In addition, localisation across multiple language sites requires extra consideration - this is based on differing sentence structure and language syntax variations between most languages.
Finally, regarding the methods
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#10
|
en most languages.
Finally, regarding the methods used, the main problem to be solved is creating a general algorithm that can recognize the entire spectrum of different voices, while disregarding nationality, gender or age. The significant differences between the extracted features - even from speakers who says the same word or phrase - must be successfully overcome.
Uses and applications
[edit]The natural-language interface gives rise to technology used for many different applications.
Some of
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#11
|
logy used for many different applications.
Some of the main uses are:
- Dictation, is the most common use for automated speech recognition (ASR) systems today. This includes medical transcriptions, legal and business dictation, and general word processing. In some cases special vocabularies are used to increase the accuracy of the system.
- Command and control, ASR systems that are designed to perform functions and actions on the system are defined as command and control systems. Utterances like
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#12
|
ed as command and control systems. Utterances like "Open Netscape" and "Start a new xterm" will do just that.
- Telephony, some PBX/Voice Mail systems allow callers to speak commands instead of pressing buttons to send specific tones.
- Wearables, because inputs are limited for wearable devices, speaking is a natural possibility.
- Medical, disabilities, many people have difficulty typing due to physical limitations such as repetitive strain injuries (RSI), muscular dystrophy, and many others. F
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#13
|
ries (RSI), muscular dystrophy, and many others. For example, people with difficulty hearing could use a system connected to their telephone to convert a caller's speech to text.
- Embedded applications, some new cellular phones include C&C speech recognition that allow utterances such as "call home". This may be a major factor in the future of automatic speech recognition and Linux.
Below are named and defined some of the applications that use natural-language recognition, and so have integrate
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#14
|
atural-language recognition, and so have integrated utilities listed above.
Ubiquity
[edit]Ubiquity, an add-on for Mozilla Firefox, is a collection of quick and easy natural-language-derived commands that act as mashups of web services, thus allowing users to get information and relate it to current and other webpages.
Wolfram Alpha
[edit]Wolfram Alpha is an online service that answers factual queries directly by computing the answer from structured data, rather than providing a list of document
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#15
|
red data, rather than providing a list of documents or web pages that might contain the answer as a search engine would.[5] It was announced in March 2009 by Stephen Wolfram, and was released to the public on May 15, 2009.[6]
Siri
[edit]Siri is an intelligent personal assistant application integrated with operating system iOS. The application uses natural language processing to answer questions and make recommendations.
Siri's marketing claims include that it adapts to a user's individual prefer
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#16
|
clude that it adapts to a user's individual preferences over time and personalizes results, and performs tasks such as making dinner reservations while trying to catch a cab.[7]
Others
[edit]- Ask.com – The original idea behind Ask Jeeves (Ask.com) was traditional keyword searching with an ability to get answers to questions posed in everyday, natural language. The current Ask.com still supports this, with added support for math, dictionary, and conversion questions.
- Braina[8] – Braina is a na
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#17
|
conversion questions.
- Braina[8] – Braina is a natural language interface for Windows OS that allows to type or speak English language sentences to perform a certain action or find information.
- GNOME Do – Allows for quick finding miscellaneous artifacts of GNOME environment (applications, Evolution and Pidgin contacts, Firefox bookmarks, Rhythmbox artists and albums, and so on) and execute the basic actions on them (launch, open, email, chat, play, etc.).[9]
- hakia – hakia was an Internet se
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#18
|
play, etc.).[9]
- hakia – hakia was an Internet search engine. The company invented an alternative new infrastructure to indexing that used SemanticRank algorithm, a solution mix from the disciplines of ontological semantics, fuzzy logic, computational linguistics, and mathematics. hakia closed in 2014.
- Lexxe – Lexxe was an Internet search engine that used natural-language processing for queries (semantic search). Searches could be made with keywords, phrases, and questions, such as "How old i
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#19
|
ywords, phrases, and questions, such as "How old is Wikipedia?" Lexxe closed its search engine services in 2015.
- Pikimal – Pikimal used natural-language tied to user preference to make search recommendations by template. Pikimal closed in 2015.
- Powerset – On May 11, 2008, the company unveiled a tool for searching a fixed subset of Wikipedia using conversational phrases rather than keywords.[10] On July 1, 2008, it was purchased by Microsoft.[11]
- Q-go – The Q-go technology provides relevant
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#20
|
11]
- Q-go – The Q-go technology provides relevant answers to users in response to queries on a company’s internet website or corporate intranet, formulated in natural sentences or keyword input alike. Q-go was acquired by RightNow Technologies in 2011.
- Yebol – Yebol was a vertical "decision" search engine that had developed a knowledge-based, semantic search platform. Yebol's artificial intelligence human intelligence-infused algorithms automatically clustered and categorized search results,
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#21
|
tically clustered and categorized search results, web sites, pages and content that it presented in a visually indexed format that is more aligned with initial human intent. Yebol used association, ranking and clustering algorithms to analyze related keywords or web pages. Yebol integrated natural-language processing, metasynthetic-engineered open complex systems, and machine algorithms with human knowledge for each query to establish a web directory that actually 'learns', using correlation, cl
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#22
|
tory that actually 'learns', using correlation, clustering and classification algorithms to automatically generate the knowledge query, which was retained and regenerated forward.[12]
See also
[edit]- Conversational user interface
- Natural user interface
- Natural-language programming
- Voice user interface
- Chatbot, a computer program that simulates human conversations
- Noisy text
- Question answering
- Selection-based search
- Semantic search
- Semantic query
- Semantic Web
References
[edit
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#23
|
h
- Semantic query
- Semantic Web
References
[edit]- ^ Hill, I. (1983). "Natural language versus computer language." In M. Sime and M. Coombs (Eds.) Designing for Human-Computer Communication. Academic Press.
- ^ a b Natural Language Interfaces to Databases – An Introduction, I. Androutsopoulos, G.D. Ritchie, P. Thanisch, Department of Artificial Intelligence, University of Edinburgh
- ^ "Chat-80 demo". Archived from the original on 11 November 2016. Retrieved 29 January 2018.
- ^ "ELIZA demo".
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#24
|
016. Retrieved 29 January 2018.
- ^ "ELIZA demo". Archived from the original on 26 November 2016. Retrieved 29 January 2018.
- ^ Johnson, Bobbie (2009-03-09). "British search engine 'could rival Google'". The Guardian. Retrieved 2009-03-09.
- ^ "So Much for A Quiet Launch". Wolfram Alpha Blog. 2009-05-08. Retrieved 2009-10-20.
- ^ "iOS - Siri". Apple. Retrieved 29 January 2018.
- ^ "Braina - Artificial Intelligence Software for Windows". www.brainasoft.com. Retrieved 29 January 2018.
- ^ Ubuntu
|
https://en.wikipedia.org/wiki/Natural-language_user_interface#25
|
nasoft.com. Retrieved 29 January 2018.
- ^ Ubuntu 10.04 Add/Remove Applications description for GNOME Do
- ^ Helft, Miguel (May 12, 2008). "Powerset Debuts With Search of Wikipedia". The New York Times.
- ^ Johnson, Mark (July 1, 2008). "Microsoft to Acquire Powerset". Powerset Blog. Archived from the original on February 25, 2009.
- ^ Humphries, Matthew. "Yebol.com steps into the search market" Archived 2012-03-15 at the Wayback Machine Geek.com. 31 July 2009.
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#0
|
BERT (language model)
Bidirectional encoder representations from transformers (BERT) is a language model introduced in October 2018 by researchers at Google.[1][2] It learns to represent text as a sequence of vectors using self-supervised learning. It uses the encoder-only transformer architecture. BERT dramatically improved the state-of-the-art for large language models. As of 2020[update], BERT is a ubiquitous baseline in natural language processing (NLP) experiments.[3]
BERT is trained by mas
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#1
|
ssing (NLP) experiments.[3]
BERT is trained by masked token prediction and next sentence prediction. As a result of this training process, BERT learns contextual, latent representations of tokens in their context, similar to ELMo and GPT-2.[4] It found applications for many natural language processing tasks, such as coreference resolution and polysemy resolution.[5] It is an evolutionary step over ELMo, and spawned the study of "BERTology", which attempts to interpret what is learned by BERT.[3]
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#2
|
attempts to interpret what is learned by BERT.[3]
BERT was originally implemented in the English language at two model sizes, BERTBASE (110 million parameters) and BERTLARGE (340 million parameters). Both were trained on the Toronto BookCorpus[6] (800M words) and English Wikipedia (2,500M words).[1]: 5 The weights were released on GitHub.[7] On March 11, 2020, 24 smaller models were released, the smallest being BERTTINY with just 4 million parameters.[7]
Architecture
[edit]BERT is an "encoder-o
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#3
|
eters.[7]
Architecture
[edit]BERT is an "encoder-only" transformer architecture. At a high level, BERT consists of 4 modules:
- Tokenizer: This module converts a piece of English text into a sequence of integers ("tokens").
- Embedding: This module converts the sequence of tokens into an array of real-valued vectors representing the tokens. It represents the conversion of discrete token types into a lower-dimensional Euclidean space.
- Encoder: a stack of Transformer blocks with self-attention,
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#4
|
stack of Transformer blocks with self-attention, but without causal masking.
- Task head: This module converts the final representation vectors into one-hot encoded tokens again by producing a predicted probability distribution over the token types. It can be viewed as a simple decoder, decoding the latent representation into token types, or as an "un-embedding layer".
The task head is necessary for pre-training, but it is often unnecessary for so-called "downstream tasks," such as question ans
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#5
|
so-called "downstream tasks," such as question answering or sentiment classification. Instead, one removes the task head and replaces it with a newly initialized module suited for the task, and finetune the new module. The latent vector representation of the model is directly fed into this new module, allowing for sample-efficient transfer learning.[1][8]
Embedding
[edit]This section describes the embedding used by BERTBASE. The other one, BERTLARGE, is similar, just larger.
The tokenizer of BER
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#6
|
RGE, is similar, just larger.
The tokenizer of BERT is WordPiece, which is a sub-word strategy like byte pair encoding. Its vocabulary size is 30,000, and any token not appearing in its vocabulary is replaced by [UNK]
("unknown").
The first layer is the embedding layer, which contains three components: token type embeddings, position embeddings, and segment type embeddings.
- Token type: The token type is a standard embedding layer, translating a one-hot vector into a dense vector based on its t
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#7
|
one-hot vector into a dense vector based on its token type.
- Position: The position embeddings are based on a token's position in the sequence. BERT uses absolute position embeddings, where each position in sequence is mapped to a real-valued vector. Each dimension of the vector consists of a sinusoidal function that takes the position in the sequence as input.
- Segment type: Using a vocabulary of just 0 or 1, this embedding layer produces a dense vector based on whether the token belongs to
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#8
|
ense vector based on whether the token belongs to the first or second text segment in that input. In other words, type-1 tokens are all tokens that appear after the
[SEP]
special token. All prior tokens are type-0.
The three embedding vectors are added together representing the initial token representation as a function of these three pieces of information. After embedding, the vector representation is normalized using a LayerNorm operation, outputting a 768-dimensional vector for each input tok
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#9
|
utting a 768-dimensional vector for each input token. After this, the representation vectors are passed forward through 12 Transformer encoder blocks, and are decoded back to 30,000-dimensional vocabulary space using a basic affine transformation layer.
Architectural family
[edit]The encoder stack of BERT has 2 free parameters: , the number of layers, and , the hidden size. There are always self-attention heads, and the feed-forward/filter size is always . By varying these two numbers, one obtai
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#10
|
s always . By varying these two numbers, one obtains an entire family of BERT models.[9]
For BERT
- the feed-forward size and filter size are synonymous. Both of them denote the number of dimensions in the middle layer of the feed-forward network.
- the hidden size and embedding size are synonymous. Both of them denote the number of real numbers used to represent a token.
The notation for encoder stack is written as L/H. For example, BERTBASE is written as 12L/768H, BERTLARGE as 24L/1024H, and B
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#11
|
written as 12L/768H, BERTLARGE as 24L/1024H, and BERTTINY as 2L/128H.
Training
[edit]Pre-training
[edit]BERT was pre-trained simultaneously on two tasks.[10]
- Masked Language Model (MLM): In this task, BERT ingests a sequence of words, where one words may be randomly changed ("masked"), and BERT tries to predict the original words that had been changed. For example, in the sentence "The cat sat on the
[MASK]
," BERT would need to predict "mat." This helps BERT learn bidirectional context, meani
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#12
|
This helps BERT learn bidirectional context, meaning it understands the relationships between words not just from left to right or right to left but from both directions at the same time.
- Next Sentence Prediction (NSP): In this task, BERT is trained to predict whether one sentence logically follows another. For example, given two sentences, "The cat sat on the mat." and "It was a sunny day," BERT has to decide if the second sentence is a valid continuation of the first one. This helps BERT und
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#13
|
continuation of the first one. This helps BERT understand relationships between sentences, which is important for tasks like question answering or document classification.
Masked language modeling
[edit]In masked language modeling, 15% of tokens would be randomly selected for masked-prediction task, and the training objective was to predict the masked token given its context. In more detail, the selected token is
- replaced with a
[MASK]
token with probability 80%, - replaced with a random word
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#14
|
th probability 80%, - replaced with a random word token with probability 10%,
- not replaced with probability 10%.
The reason not all selected tokens are masked is to avoid the dataset shift problem. The dataset shift problem arises when the distribution of inputs seen during training differs significantly from the distribution encountered during inference. A trained BERT model might be applied to word representation (like Word2Vec), where it would be run over sentences not containing any [MASK]
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#15
|
ld be run over sentences not containing any [MASK]
tokens. It is later found that more diverse training objectives are generally better.[11]
As an illustrative example, consider the sentence "my dog is cute". It would first be divided into tokens like "my1 dog2 is3 cute4". Then a random token in the sentence would be picked. Let it be the 4th one "cute4". Next, there would be three possibilities:
- with probability 80%, the chosen token is masked, resulting in "my1 dog2 is3
[MASK]
4"; - with pro
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#16
|
, resulting in "my1 dog2 is3
[MASK]
4"; - with probability 10%, the chosen token is replaced by a uniformly sampled random token, such as "happy", resulting in "my1 dog2 is3 happy4";
- with probability 10%, nothing is done, resulting in "my1 dog2 is3 cute4".
After processing the input text, the model's 4th output vector is passed to its decoder layer, which outputs a probability distribution over its 30,000-dimensional vocabulary space.
Next sentence prediction
[edit]Given two spans of text, the
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#17
|
ence prediction
[edit]Given two spans of text, the model predicts if these two spans appeared sequentially in the training corpus, outputting either [IsNext]
or [NotNext]
. Specifically, the training algorithm would sometimes sample two spans from a single continuous span in the training corpus, but other times, sample two spans from two discontinuous spans in the training corpus.
The first span starts with a special token [CLS]
(for "classify"). The two spans are separated by a special token [S
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#18
|
The two spans are separated by a special token [SEP]
(for "separate"). After processing the two spans, the 1-st output vector (the vector coding for [CLS]
) is passed to a separate neural network for the binary classification into [IsNext]
and [NotNext]
.
- For example, given "
[CLS]
my dog is cute[SEP]
he likes playing" the model should output token[IsNext]
. - Given "
[CLS]
my dog is cute[SEP]
how do magnets work" the model should output token[NotNext]
.
Fine-tuning
[edit]-
Sentiment classifi
|
https://en.wikipedia.org/wiki/BERT_%28language_model%29#19
|
[NotNext]
.
Fine-tuning
[edit]-
Sentiment classification
-
Sentence classification
-
Answering multiple-choice questions
BERT is meant as a general pretrained model for various applications in natural language processing. That is, after pre-training, BERT can be fine-tuned with fewer resources on smaller datasets to optimize its performance on specific tasks such as natural language inference and text classification, and sequence-to-sequence-based language generation tasks such as question answe
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.