| { | |
| "architectures": [ | |
| "TinyIndianLMForCausalLM" | |
| ], | |
| "bos_token_id": 1, | |
| "d_ffn": 1024, | |
| "d_model": 384, | |
| "dropout": 0.1, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "experiment_family": "Monolingual \u2014 Indo-Aryan", | |
| "experiment_langs": [ | |
| "hindi" | |
| ], | |
| "experiment_name": "hi", | |
| "lang_token_ids": { | |
| "bangla": 5, | |
| "hindi": 4, | |
| "kannada": 7, | |
| "malayalam": 6 | |
| }, | |
| "max_seq_len": 256, | |
| "model_type": "tinyindianlm", | |
| "n_heads": 8, | |
| "n_kv_heads": null, | |
| "n_layers": 8, | |
| "pad_token_id": 0, | |
| "transformers_version": "5.0.0", | |
| "val_ppl": 3.88, | |
| "vocab_size": 12088 | |
| } | |