PredictiveManish/Trimurti-LM

1{2	"model_id": "trimurti-lm",3	"model_name": "Trimurti-LM",4	"model_type": "language_model",5	"architecture": "GPT2",6	"framework": "transformers",7	8	"languages": {9		"supported": ["en", "hi", "pa"],10		"language_tags": ["[EN]", "[HI]", "[PA]"],11		"description": "Trilingual language model supporting English, Hindi, and Punjabi"12	},13	14	"model_config": {15		"vocab_size": 8000,16		"n_positions": 128,17		"n_embd": 256,18		"n_layer": 4,19		"n_head": 4,20		"n_inner": 512,21		"activation_function": "gelu_new",22		"attn_pdrop": 0.1,23		"embd_pdrop": 0.1,24		"resid_pdrop": 0.1,25		"estimated_parameters": "4.7M"26	},27	28	"tokenizer": {29		"type": "sentencepiece",30		"model_type": "unigram",31		"vocab_size": 8000,32		"character_coverage": 0.9995,33		"byte_fallback": true,34		"model_path": "final_corpus/multilingual_spm.model"35	},36	37	"training": {38		"corpus": "final_corpus/multilingual_corpus_train.txt",39		"validation": "final_corpus/multilingual_corpus_val.txt",40		"total_steps": 5000,41		"batch_size": 2,42		"gradient_accumulation": 8,43		"learning_rate": 2e-4,44		"warmup_steps": 1000,45		"effective_batch_size": 1646	},47	48	"checkpoints": {49		"path": "checkpoints_tiny",50		"available_checkpoints": [51			"step1000",52			"step2000",53			"step3000",54			"step4000",55			"step5000",56			"final"57		]58	},59	60	"evaluation": {61		"overall_accuracy": 100.0,62		"english_accuracy": 100.0,63		"hindi_accuracy": 100.0,64		"punjabi_accuracy": 100.0,65		"mixed_accuracy": 100.0,66		"avg_english_perplexity": 42.29,67		"avg_hindi_perplexity": 50.56,68		"avg_punjabi_perplexity": 63.4269	},70	71	"entry_points": {72		"training": "python train_model.py",73		"testing": "python test_model.py",74		"evaluation": "python evaluate_model.py",75		"preprocessing": "python preprocess.py",76		"web_interface": "python web_interface.py"77	},78	79	"dependencies": [80		"torch",81		"transformers",82		"sentencepiece",83		"tqdm",84		"gradio",85		"pandas",86		"numpy"87	],88	89	"filter": [90		{91			"bool": {92				"should": [93					{94						"term": { "path": "model_index.json" }95					},96					{97						"regexp": { "path": "[^/]*\\.safetensors" }98					},99					{100						"regexp": { "path": "[^/]*\\.ckpt" }101					},102					{103						"regexp": { "path": "[^/]*\\.bin" }104					}105				],106				"minimum_should_match": 1107			}108		}109	],110	111	"files": [112		{113			"path": "checkpoints_tiny/final/model.safetensors",114			"description": "Final trained model weights"115		},116		{117			"path": "checkpoints_tiny/final/config.json",118			"description": "Model configuration"119		},120		{121			"path": "checkpoints_tiny/final/generation_config.json",122			"description": "Generation settings"123		},124		{125			"path": "final_corpus/multilingual_spm.model",126			"description": "SentencePiece tokenizer model"127		},128		{129			"path": "final_corpus/multilingual_spm.vocab",130			"description": "Tokenizer vocabulary"131		},132		{133			"path": "train_model.py",134			"description": "Training script"135		},136		{137			"path": "test_model.py",138			"description": "Testing and inference script"139		},140		{141			"path": "evaluate_model.py",142			"description": "Evaluation script"143		},144		{145			"path": "preprocess.py",146			"description": "Data preprocessing script"147		},148		{149			"path": "web_interface.py",150			"description": "Gradio web interface"151		}152	]153}