Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Very Low Activity

Commits : Listings

Analyzed about 2 hours ago. based on code collected 1 day ago.
Nov 05, 2025 — Nov 05, 2026
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Updated Requirements.txt
K.Srinivasan
More... over 1 year ago
Section 1579 updated
K.Srinivasan
More... almost 2 years ago
Meta AI perplexity score for Section 1579
K.Srinivasan
More... almost 2 years ago
Section 1579 updated
K.Srinivasan
More... almost 2 years ago
Section 1553 updated
K.Srinivasan
More... almost 2 years ago
Section 1553 updated
K.Srinivasan
More... almost 2 years ago
Section 1579 updated
K.Srinivasan
More... almost 2 years ago
Section 1579 updated
Srinivasan Kannan
More... almost 2 years ago
Section 1533 updated
Srinivasan Kannan
More... almost 2 years ago
Section 1579 updated
Srinivasan Kannan
More... almost 2 years ago
1579. (THEORY and FEATURE) Nachiketas Formal LLM - Part-of-speech tagged_words() from NLTK and user-defined corpus - Changes to Wikipedia Question-Answering algorithm for better accuracy 17 - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression - 2 February 2025
Srinivasan Kannan
More... almost 2 years ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1579. (THEORY and FEATURE) Nachiketas Formal LLM - Part-of-speech tagged_words() from NLTK and user-defined corpus - Changes to Wikipedia Question-Answering algorithm for better accuracy 17 - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576,1578 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression - 2 February 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1579.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import more corpora - brown and conll2000 - from NLTK corpus module. 1579.2 Function def WikipediaRLFGTransformersQuestionAnswering(question,questionfraction=1,maxanswers=1,keywordsearch=False,wsheading=True,answerslice=1,answerfraction=1,bothvertices_intersection=True,sentence_type="xtag_node34_triplets",number_of_random_walks=10,number_of_words_per_sentence=5,number_of_cores_per_random_walk=5,std_sentence_PoS_dict={"ADJ":[],"PROPN":[],"NOUN":[],"AUX":[],"ADP":[],"ADV":[],"VERB":[],"DET":[],"PRON":[],"CCONJ":[],"NUM":[],"SYM":[],"X":[]},blanks=False,perplexity_algorithm="WordNet",treenode_type="PoS",sentence_tuple_array=False,sentence_PoS_array=None,randomwalk_to_sentence_template_ratio=10,user_defined_PoS2Vocabulary_dict=None) has been enhanced to add argument user_defined_PoSVocabulary_dict through which a dictionary mapping Part-of-speech to list of words for that Part-of-speech can be passed on by user either manually or from predefined PoS tagged_words in NLTK corpus. This dictionary mapping is percolated down to make_sentence2(). So far PoS tagged words were restricted to Penn Treebank WSJ corpus. 1579.3 Function def make_sentence2(randomwalkvertices,sentence_PoS_tuple_array=[],treenode_type="tag",max_synth_sentences=1000,user_defined_PoS2Vocabulary_dict=None) has been changed to add argument user_defined_PoS2Vocabulary_dict set default to None. 1579.4 Additional clause has been introduced for checking if user_defined_PoS2Vocabulary_dict is not Null and looking it up to get the list of words from user defined corpus for an unfilled Part-of-speech in sentence template. 1579.5 An example NLTK corpus conll2000 has been imported to get tagged_words() PoS-words dictionary by ConditionalFreqDist. Function WikipediaRLFGTransformersQuestionAnswering() is invoked twice for dictionary sentence template and array sentence template for user-defined conll2000 PoS-to-vocabulary dictionary: WikipediaRLFGTransformersQuestionAnswering(question,wsheading=True,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=50,std_sentence_PoS_dict=sentencePoSdict,number_of_cores_per_random_walk=3,number_of_random_walks=3,blanks=False,treenode_type="tag",user_defined_PoS2Vocabulary_dict=conll2000_corpus_PoS2Vocabulary_dict) ... WikipediaRLFGTransformersQuestionAnswering(question,wsheading=True,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=10,std_sentence_PoS_dict={},number_of_cores_per_random_walk=5,number_of_random_walks=5,blanks=False,treenode_type="tag",sentence_tuple_array=True, sentence_PoS_array=sentencePoSarray, randomwalk_to_sentence_template_ratio=3,user_defined_PoS2Vocabulary_dict=conll2000_corpus_PoS2Vocabulary_dict) 1579.6 Logs for an example query - python3.9 QuestionAnswering.py "define prime numbers" - are in testlogs/QuestionAnswering.log.2February20251579.7 This code change is quite important as it has provisions for flexible fine tuning of synthesized sentences which can be enriched for their meaning (perplexity) by choosing a corpus tagged_words which has closest relevance to the question - For example, earlier question on prime numbers must correspond to a corpus predefined for mathematics. Perplexity of ~2.5 attained by Nachiketas Formal LLM thus far could be made closer to 1 (i.e LLM generated sentence is as good as human sentence) through appropriate choice of user-defined PoS-to-Vocabulary tagged_words which could also be manually defined without depending on NLTK corpus - for instance by partitioning Oxford dictionary vocabulary by their PoS names and creating a PoS-Vocabulary mapping). 1579.8 Brown corpus was tried but there is a mismatch in PoS naming convention between Penn Treebank and Brown causing errors.
Srinivasan Kannan
More... almost 2 years ago
Updated BestPractices.txt
K.Srinivasan
More... almost 2 years ago
Updated index.rst
K.Srinivasan
More... almost 2 years ago
Section 1578 updated
K.Srinivasan
More... almost 2 years ago
Meta AI Llama LLM perplexity scores for: (1) Chomsky sentence (2) Nachiketas Formal LLM answer sentence before PoS refinement - QuestionAnswering.log.12January2025 (3) 3 Nachiketas Formal LLM answer sentences after PoS refinement - QuestionAnswering.log.21January2025.xz
K.Srinivasan
More... almost 2 years ago
Section 1533 updated
K.Srinivasan
More... almost 2 years ago
Section 1577 updated
K.Srinivasan
More... almost 2 years ago
Section 1578 updated
K.Srinivasan
More... almost 2 years ago
Section 1533 updated
K.Srinivasan
More... almost 2 years ago
Section 1533 updated
K.Srinivasan
More... almost 2 years ago
Section 1578 updated
K.Srinivasan
More... almost 2 years ago
Section 1488 updated
Srinivasan Kannan
More... almost 2 years ago
+1578. (THEORY and FEATURE) Nachiketas Formal LLM - Changes to Penn Treebank datasets, comparison of dictionary sentence template and array sentence template and some additional tuning - Changes to Wikipedia Question-Answering algorithm for better accuracy 16 - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression - 21 January 2025
Srinivasan Kannan
More... almost 2 years ago
-------------------------------------------------------------------------------------------------------------------------------------------- 1578. (THEORY and FEATURE) Nachiketas Formal LLM - Changes to Penn Treebank datasets, comparison of dictionary sentence template and array sentence template and some additional tuning - Changes to Wikipedia Question-Answering algorithm for better accuracy 16 - related to 2,249,707,829,1133,1134,1204,1263,1269,1284,1285,1405,1413,1455,1468,1469,1536,1537,1538,1539,1540,1541,1542,1543,1550,1568,1572,1573,1574,1575,1576 and all sections on Intrinsic Merit of Texts, Recursive Lambda Function Growth and Recursive Gloss Overlap textgraph algorithms, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains, Large Language Models, Mildly Context Sensitive Grammars-Tree Adjoining Grammars for Natural Languages,Text restoration-Vowelless text (de)compression - 21 January 2025 -------------------------------------------------------------------------------------------------------------------------------------------- 1578.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to change datatypes of perplexity dictionaries for synthesized sentences to defaultdict(list) from defaultdict(int) as more than one sentence could have same perplexity score. 1578.2 Non-PoS tagnames in sentences synthesized from sentence templates are disallowed and coded by "---" string signifying blank. 1578.3 Sentences synthesized by dictionary templates and array sentences are compared for two Penn Treebank datasets wsj_0002.mrg and wsj_0004.mrg. 1578.4 WikipediaRLFGTransformersQuestionAnswering() is invoked for increased values of cores in random walks, more random walks and shorter sentences: WikipediaRLFGTransformersQuestionAnswering(question,wsheading=True,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=10,std_sentence_PoS_dict={},number_of_cores_per_random_walk=5,number_of_random_walks=5,blanks=False,treenode_type="tag",sentence_tuple_array=True, sentence_PoS_array=sentencePoSarray, randomwalk_to_sentence_template_ratio=3) 1578.5 An example query - python3.9 QuestionAnswering.py "define unique factorization of an integer" - has been logged to testlogs/QuestionAnswering.log.21January2025.xz (compressed)
Srinivasan Kannan
More... almost 2 years ago
Section 1576 updated
K.Srinivasan
More... almost 2 years ago
Section 1576 updated
K.Srinivasan
More... almost 2 years ago
Section 1560 updated
K.Srinivasan
More... almost 2 years ago
Section 1175 updated
K.Srinivasan
More... almost 2 years ago
Section 1175 updated
K.Srinivasan
More... almost 2 years ago