Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

added_tokens.json

@@ -0,0 +1,901 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|alias_py -> identifier_py |>": 152363,
+  "<|aliased_import_py -> name_py as_ter alias_py |>": 151793,
+  "<|alternative_py -> elif_clause_py |>": 151686,
+  "<|alternative_py -> else_clause_py |>": 151695,
+  "<|argument_list_py -> (_ter|>": 152255,
+  "<|argument_list_py -> )_ter|>": 152514,
+  "<|argument_list_py -> ,_ter|>": 152309,
+  "<|argument_list_py -> End |>": 152485,
+  "<|argument_list_py -> attribute_py|>": 152280,
+  "<|argument_list_py -> binary_operator_py|>": 151837,
+  "<|argument_list_py -> boolean_operator_py|>": 152279,
+  "<|argument_list_py -> call_py|>": 151699,
+  "<|argument_list_py -> comparison_operator_py|>": 152495,
+  "<|argument_list_py -> conditional_expression_py|>": 152452,
+  "<|argument_list_py -> dictionary_comprehension_py|>": 152346,
+  "<|argument_list_py -> dictionary_py|>": 151970,
+  "<|argument_list_py -> false_py|>": 151897,
+  "<|argument_list_py -> float_py|>": 152269,
+  "<|argument_list_py -> generator_expression_py|>": 152369,
+  "<|argument_list_py -> identifier_py|>": 151884,
+  "<|argument_list_py -> identifier_ter|>": 152359,
+  "<|argument_list_py -> integer_py|>": 151949,
+  "<|argument_list_py -> keyword_argument_py|>": 152122,
+  "<|argument_list_py -> lambda_py|>": 152428,
+  "<|argument_list_py -> list_comprehension_py|>": 152347,
+  "<|argument_list_py -> list_py|>": 152060,
+  "<|argument_list_py -> list_splat_py|>": 152066,
+  "<|argument_list_py -> none_py|>": 152203,
+  "<|argument_list_py -> not_operator_py|>": 152230,
+  "<|argument_list_py -> parenthesized_expression_py|>": 152047,
+  "<|argument_list_py -> set_comprehension_py|>": 152464,
+  "<|argument_list_py -> set_py|>": 152007,
+  "<|argument_list_py -> string_literal_py|>": 152486,
+  "<|argument_list_py -> subscript_py|>": 151995,
+  "<|argument_list_py -> true_py|>": 151992,
+  "<|argument_list_py -> tuple_py|>": 152517,
+  "<|argument_list_py -> unary_operator_py|>": 151886,
+  "<|argument_py -> attribute_py |>": 151745,
+  "<|argument_py -> binary_operator_py |>": 152223,
+  "<|argument_py -> call_py |>": 152430,
+  "<|argument_py -> comparison_operator_py |>": 151876,
+  "<|argument_py -> dictionary_py |>": 152393,
+  "<|argument_py -> false_py |>": 152408,
+  "<|argument_py -> float_py |>": 152454,
+  "<|argument_py -> generator_expression_py |>": 152107,
+  "<|argument_py -> identifier_py |>": 151673,
+  "<|argument_py -> identifier_ter |>": 152341,
+  "<|argument_py -> integer_py |>": 151762,
+  "<|argument_py -> list_comprehension_py |>": 152385,
+  "<|argument_py -> list_py |>": 152054,
+  "<|argument_py -> not_operator_py |>": 152496,
+  "<|argument_py -> parenthesized_expression_py |>": 151674,
+  "<|argument_py -> string_literal_py |>": 152379,
+  "<|argument_py -> subscript_py |>": 152182,
+  "<|argument_py -> true_py |>": 151961,
+  "<|argument_py -> tuple_py |>": 152259,
+  "<|argument_py -> unary_operator_py |>": 152232,
+  "<|arguments_py -> argument_list_py |>": 152354,
+  "<|arguments_py -> generator_expression_py |>": 152478,
+  "<|assert_statement_py -> ,_ter|>": 152145,
+  "<|assert_statement_py -> End |>": 152407,
+  "<|assert_statement_py -> assert_ter|>": 152113,
+  "<|assert_statement_py -> binary_operator_py|>": 151742,
+  "<|assert_statement_py -> boolean_operator_py|>": 151831,
+  "<|assert_statement_py -> call_py|>": 152515,
+  "<|assert_statement_py -> comparison_operator_py|>": 151722,
+  "<|assert_statement_py -> generator_expression_py|>": 152318,
+  "<|assert_statement_py -> identifier_py|>": 152300,
+  "<|assert_statement_py -> parenthesized_expression_py|>": 151936,
+  "<|assert_statement_py -> string_literal_py|>": 151717,
+  "<|assert_statement_py -> tuple_py|>": 152356,
+  "<|assignment_py -> left_py :_ter type_py =_ter right_py |>": 152011,
+  "<|assignment_py -> left_py :_ter type_py |>": 151679,
+  "<|assignment_py -> left_py =_ter right_py |>": 151846,
+  "<|attribute_py -> identifier_py |>": 151994,
+  "<|attribute_py -> identifier_ter |>": 151900,
+  "<|attribute_py -> object_py ._ter attribute_py |>": 152340,
+  "<|augmented_assignment_py -> left_py operator_py right_py |>": 151963,
+  "<|binary_operator_py -> left_py operator_py right_py |>": 151931,
+  "<|block_py -> ;_ter|>": 152271,
+  "<|block_py -> End |>": 151953,
+  "<|block_py -> assert_statement_py|>": 151797,
+  "<|block_py -> break_statement_py|>": 151895,
+  "<|block_py -> class_definition_py|>": 152109,
+  "<|block_py -> continue_statement_py|>": 152421,
+  "<|block_py -> delete_statement_py|>": 152261,
+  "<|block_py -> expression_statement_py|>": 151719,
+  "<|block_py -> for_statement_py|>": 152026,
+  "<|block_py -> function_definition_py|>": 152512,
+  "<|block_py -> future_import_statement_py|>": 152059,
+  "<|block_py -> if_statement_py|>": 152115,
+  "<|block_py -> import_from_statement_py|>": 151916,
+  "<|block_py -> import_statement_py|>": 152033,
+  "<|block_py -> pass_statement_py|>": 152048,
+  "<|block_py -> print_statement_py|>": 151791,
+  "<|block_py -> raise_statement_py|>": 152466,
+  "<|block_py -> return_statement_py|>": 151800,
+  "<|block_py -> try_statement_py|>": 152187,
+  "<|block_py -> type_alias_statement_py|>": 152388,
+  "<|block_py -> while_statement_py|>": 152527,
+  "<|body_py -> End |>": 152146,
+  "<|body_py -> attribute_py|>": 152268,
+  "<|body_py -> binary_operator_py|>": 152178,
+  "<|body_py -> block_py|>": 152498,
+  "<|body_py -> boolean_operator_py|>": 152311,
+  "<|body_py -> call_py|>": 152156,
+  "<|body_py -> comment_py|>": 151739,
+  "<|body_py -> comparison_operator_py|>": 152457,
+  "<|body_py -> conditional_expression_py|>": 151951,
+  "<|body_py -> dictionary_comprehension_py|>": 152319,
+  "<|body_py -> dictionary_py|>": 151902,
+  "<|body_py -> false_py|>": 152465,
+  "<|body_py -> float_py|>": 151729,
+  "<|body_py -> generator_expression_py|>": 152351,
+  "<|body_py -> identifier_py|>": 151962,
+  "<|body_py -> identifier_ter|>": 151955,
+  "<|body_py -> integer_py|>": 152003,
+  "<|body_py -> lambda_py|>": 151852,
+  "<|body_py -> list_comprehension_py|>": 152112,
+  "<|body_py -> list_py|>": 151740,
+  "<|body_py -> none_py|>": 152307,
+  "<|body_py -> not_operator_py|>": 152487,
+  "<|body_py -> pair_py|>": 152151,
+  "<|body_py -> parenthesized_expression_py|>": 152260,
+  "<|body_py -> set_comprehension_py|>": 152438,
+  "<|body_py -> set_py|>": 152221,
+  "<|body_py -> string_literal_py|>": 151712,
+  "<|body_py -> subscript_py|>": 152104,
+  "<|body_py -> true_py|>": 152045,
+  "<|body_py -> tuple_py|>": 152022,
+  "<|body_py -> unary_operator_py|>": 151882,
+  "<|boolean_operator_py -> left_py operator_py right_py |>": 151968,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|break_statement_py -> break_ter |>": 152426,
+  "<|call_py -> function_py arguments_py |>": 152165,
+  "<|chevron_py -> >>_ter attribute_py |>": 152315,
+  "<|chevron_py -> >>_ter identifier_py |>": 151758,
+  "<|class_definition_py -> class_ter name_py :_ter body_py |>": 152086,
+  "<|class_definition_py -> class_ter name_py superclasses_py :_ter body_py |>": 152509,
+  "<|comment_py -> End|>": 152488,
+  "<|comparison_operator_py -> End |>": 152088,
+  "<|comparison_operator_py -> attribute_py|>": 152164,
+  "<|comparison_operator_py -> binary_operator_py|>": 152322,
+  "<|comparison_operator_py -> call_py|>": 152158,
+  "<|comparison_operator_py -> dictionary_comprehension_py|>": 152390,
+  "<|comparison_operator_py -> dictionary_py|>": 152035,
+  "<|comparison_operator_py -> false_py|>": 151794,
+  "<|comparison_operator_py -> float_py|>": 151703,
+  "<|comparison_operator_py -> generator_expression_py|>": 152314,
+  "<|comparison_operator_py -> identifier_py|>": 152327,
+  "<|comparison_operator_py -> identifier_ter|>": 152027,
+  "<|comparison_operator_py -> integer_py|>": 152447,
+  "<|comparison_operator_py -> list_comprehension_py|>": 151878,
+  "<|comparison_operator_py -> list_py|>": 151912,
+  "<|comparison_operator_py -> list_splat_py|>": 152192,
+  "<|comparison_operator_py -> none_py|>": 152111,
+  "<|comparison_operator_py -> operators_py|>": 151733,
+  "<|comparison_operator_py -> parenthesized_expression_py|>": 152326,
+  "<|comparison_operator_py -> set_comprehension_py|>": 152533,
+  "<|comparison_operator_py -> set_py|>": 152422,
+  "<|comparison_operator_py -> string_literal_py|>": 152387,
+  "<|comparison_operator_py -> subscript_py|>": 151697,
+  "<|comparison_operator_py -> true_py|>": 151795,
+  "<|comparison_operator_py -> tuple_py|>": 152436,
+  "<|comparison_operator_py -> unary_operator_py|>": 152431,
+  "<|condition_py -> attribute_py |>": 152446,
+  "<|condition_py -> binary_operator_py |>": 152423,
+  "<|condition_py -> boolean_operator_py |>": 152120,
+  "<|condition_py -> call_py |>": 152250,
+  "<|condition_py -> comparison_operator_py |>": 152205,
+  "<|condition_py -> conditional_expression_py |>": 152002,
+  "<|condition_py -> false_py |>": 152074,
+  "<|condition_py -> identifier_py |>": 152275,
+  "<|condition_py -> identifier_ter |>": 152247,
+  "<|condition_py -> integer_py |>": 151820,
+  "<|condition_py -> not_operator_py |>": 151930,
+  "<|condition_py -> parenthesized_expression_py |>": 152175,
+  "<|condition_py -> string_literal_py |>": 152153,
+  "<|condition_py -> subscript_py |>": 152038,
+  "<|condition_py -> true_py |>": 152306,
+  "<|condition_py -> tuple_py |>": 151971,
+  "<|condition_py -> unary_operator_py |>": 152456,
+  "<|conditional_expression_py -> End |>": 152089,
+  "<|conditional_expression_py -> attribute_py|>": 152181,
+  "<|conditional_expression_py -> binary_operator_py|>": 152061,
+  "<|conditional_expression_py -> boolean_operator_py|>": 152276,
+  "<|conditional_expression_py -> call_py|>": 151746,
+  "<|conditional_expression_py -> comparison_operator_py|>": 152095,
+  "<|conditional_expression_py -> conditional_expression_py|>": 151973,
+  "<|conditional_expression_py -> dictionary_comprehension_py|>": 151670,
+  "<|conditional_expression_py -> dictionary_py|>": 152405,
+  "<|conditional_expression_py -> else_ter|>": 151947,
+  "<|conditional_expression_py -> false_py|>": 152461,
+  "<|conditional_expression_py -> float_py|>": 152133,
+  "<|conditional_expression_py -> generator_expression_py|>": 152125,
+  "<|conditional_expression_py -> identifier_py|>": 151965,
+  "<|conditional_expression_py -> identifier_ter|>": 152023,
+  "<|conditional_expression_py -> if_ter|>": 152069,
+  "<|conditional_expression_py -> integer_py|>": 152106,
+  "<|conditional_expression_py -> lambda_py|>": 152000,
+  "<|conditional_expression_py -> list_comprehension_py|>": 152521,
+  "<|conditional_expression_py -> list_py|>": 152208,
+  "<|conditional_expression_py -> none_py|>": 152019,
+  "<|conditional_expression_py -> not_operator_py|>": 152373,
+  "<|conditional_expression_py -> parenthesized_expression_py|>": 152087,
+  "<|conditional_expression_py -> set_comprehension_py|>": 152229,
+  "<|conditional_expression_py -> set_py|>": 152516,
+  "<|conditional_expression_py -> string_literal_py|>": 152206,
+  "<|conditional_expression_py -> subscript_py|>": 152523,
+  "<|conditional_expression_py -> true_py|>": 151692,
+  "<|conditional_expression_py -> tuple_py|>": 152350,
+  "<|conditional_expression_py -> unary_operator_py|>": 151803,
+  "<|consequence_py -> block_py |>": 151972,
+  "<|consequence_py -> comment_py |>": 152480,
+  "<|continue_statement_py -> continue_ter |>": 151941,
+  "<|default_parameter_py -> name_py =_ter value_py |>": 152325,
+  "<|delete_statement_py -> del_ter attribute_py |>": 152449,
+  "<|delete_statement_py -> del_ter binary_operator_py |>": 152284,
+  "<|delete_statement_py -> del_ter call_py |>": 151868,
+  "<|delete_statement_py -> del_ter expression_list_py |>": 152030,
+  "<|delete_statement_py -> del_ter identifier_py |>": 152236,
+  "<|delete_statement_py -> del_ter parenthesized_expression_py |>": 152052,
+  "<|delete_statement_py -> del_ter subscript_py |>": 152162,
+  "<|delete_statement_py -> del_ter true_py |>": 151857,
+  "<|dictionary_comprehension_py -> End |>": 152036,
+  "<|dictionary_comprehension_py -> body_py|>": 152287,
+  "<|dictionary_comprehension_py -> for_in_clause_py|>": 151989,
+  "<|dictionary_comprehension_py -> if_clause_py|>": 151806,
+  "<|dictionary_comprehension_py -> {_ter|>": 152043,
+  "<|dictionary_comprehension_py -> }_ter|>": 152282,
+  "<|dictionary_py -> ,_ter|>": 151816,
+  "<|dictionary_py -> End |>": 151668,
+  "<|dictionary_py -> pair_py|>": 151779,
+  "<|dictionary_py -> {_ter|>": 152154,
+  "<|dictionary_py -> }_ter|>": 152286,
+  "<|dotted_name_py -> End |>": 151775,
+  "<|dotted_name_py -> identifier_py|>": 151996,
+  "<|elif_clause_py -> elif_ter condition_py :_ter consequence_py |>": 151922,
+  "<|else_clause_py -> else_ter :_ter body_py |>": 152179,
+  "<|endoftext|>": 151643,
+  "<|except_clause_py -> except_ter :_ter block_py |>": 152335,
+  "<|except_clause_py -> except_ter value_py :_ter block_py |>": 151952,
+  "<|except_clause_py -> except_ter value_py :_ter |>": 152333,
+  "<|expression_list_py -> ,_ter|>": 151889,
+  "<|expression_list_py -> End |>": 152411,
+  "<|expression_list_py -> attribute_py|>": 151804,
+  "<|expression_list_py -> binary_operator_py|>": 152281,
+  "<|expression_list_py -> boolean_operator_py|>": 152310,
+  "<|expression_list_py -> call_py|>": 151817,
+  "<|expression_list_py -> comparison_operator_py|>": 152502,
+  "<|expression_list_py -> conditional_expression_py|>": 151986,
+  "<|expression_list_py -> dictionary_comprehension_py|>": 151859,
+  "<|expression_list_py -> dictionary_py|>": 151901,
+  "<|expression_list_py -> false_py|>": 152009,
+  "<|expression_list_py -> float_py|>": 151755,
+  "<|expression_list_py -> generator_expression_py|>": 151678,
+  "<|expression_list_py -> identifier_py|>": 152510,
+  "<|expression_list_py -> identifier_ter|>": 152090,
+  "<|expression_list_py -> integer_py|>": 152083,
+  "<|expression_list_py -> lambda_py|>": 152256,
+  "<|expression_list_py -> list_comprehension_py|>": 152049,
+  "<|expression_list_py -> list_py|>": 151858,
+  "<|expression_list_py -> list_splat_py|>": 152470,
+  "<|expression_list_py -> none_py|>": 151959,
+  "<|expression_list_py -> not_operator_py|>": 152099,
+  "<|expression_list_py -> parenthesized_expression_py|>": 151908,
+  "<|expression_list_py -> set_comprehension_py|>": 152163,
+  "<|expression_list_py -> set_py|>": 152479,
+  "<|expression_list_py -> string_literal_py|>": 152372,
+  "<|expression_list_py -> subscript_py|>": 151736,
+  "<|expression_list_py -> true_py|>": 151975,
+  "<|expression_list_py -> tuple_py|>": 152475,
+  "<|expression_list_py -> unary_operator_py|>": 152317,
+  "<|expression_statement_py -> ,_ter|>": 151862,
+  "<|expression_statement_py -> End |>": 151983,
+  "<|expression_statement_py -> assignment_py|>": 152176,
+  "<|expression_statement_py -> attribute_py|>": 151777,
+  "<|expression_statement_py -> augmented_assignment_py|>": 152068,
+  "<|expression_statement_py -> binary_operator_py|>": 152357,
+  "<|expression_statement_py -> boolean_operator_py|>": 152183,
+  "<|expression_statement_py -> call_py|>": 151958,
+  "<|expression_statement_py -> comparison_operator_py|>": 151693,
+  "<|expression_statement_py -> conditional_expression_py|>": 152348,
+  "<|expression_statement_py -> dictionary_comprehension_py|>": 152147,
+  "<|expression_statement_py -> dictionary_py|>": 151938,
+  "<|expression_statement_py -> false_py|>": 151906,
+  "<|expression_statement_py -> float_py|>": 151763,
+  "<|expression_statement_py -> generator_expression_py|>": 151737,
+  "<|expression_statement_py -> identifier_py|>": 152248,
+  "<|expression_statement_py -> integer_py|>": 151799,
+  "<|expression_statement_py -> lambda_py|>": 152044,
+  "<|expression_statement_py -> list_comprehension_py|>": 152471,
+  "<|expression_statement_py -> list_py|>": 152453,
+  "<|expression_statement_py -> none_py|>": 152097,
+  "<|expression_statement_py -> not_operator_py|>": 151913,
+  "<|expression_statement_py -> parenthesized_expression_py|>": 151910,
+  "<|expression_statement_py -> set_comprehension_py|>": 152384,
+  "<|expression_statement_py -> set_py|>": 152219,
+  "<|expression_statement_py -> string_literal_py|>": 152239,
+  "<|expression_statement_py -> subscript_py|>": 152530,
+  "<|expression_statement_py -> true_py|>": 151866,
+  "<|expression_statement_py -> tuple_py|>": 152312,
+  "<|expression_statement_py -> unary_operator_py|>": 151821,
+  "<|expression_statement_py -> yield_py|>": 151770,
+  "<|false_py -> False_ter |>": 151818,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|for_in_clause_py -> End |>": 151933,
+  "<|for_in_clause_py -> for_ter|>": 151805,
+  "<|for_in_clause_py -> in_ter|>": 152212,
+  "<|for_in_clause_py -> left_py|>": 151915,
+  "<|for_in_clause_py -> right_py|>": 151724,
+  "<|for_statement_py -> for_ter left_py in_ter right_py :_ter body_py alternative_py |>": 151999,
+  "<|for_statement_py -> for_ter left_py in_ter right_py :_ter body_py |>": 152063,
+  "<|function_definition_py -> def_ter name_py parameters_py ->_ter return_type_py :_ter body_py |>": 152039,
+  "<|function_definition_py -> def_ter name_py parameters_py :_ter body_py |>": 152291,
+  "<|function_py -> attribute_py |>": 152228,
+  "<|function_py -> call_py |>": 151730,
+  "<|function_py -> false_py |>": 152491,
+  "<|function_py -> identifier_py |>": 151759,
+  "<|function_py -> identifier_ter |>": 151891,
+  "<|function_py -> integer_py |>": 152222,
+  "<|function_py -> list_splat_py |>": 152102,
+  "<|function_py -> none_py |>": 152184,
+  "<|function_py -> parenthesized_expression_py |>": 151750,
+  "<|function_py -> subscript_py |>": 151824,
+  "<|function_py -> tuple_py |>": 151721,
+  "<|future_import_statement_py -> ,_ter|>": 151835,
+  "<|future_import_statement_py -> End |>": 152361,
+  "<|future_import_statement_py -> __future___ter|>": 152448,
+  "<|future_import_statement_py -> from_ter|>": 151801,
+  "<|future_import_statement_py -> import_ter|>": 151760,
+  "<|future_import_statement_py -> name_py|>": 152034,
+  "<|generator_expression_py -> (_ter|>": 152355,
+  "<|generator_expression_py -> )_ter|>": 152072,
+  "<|generator_expression_py -> End |>": 152492,
+  "<|generator_expression_py -> body_py|>": 151950,
+  "<|generator_expression_py -> for_in_clause_py|>": 152420,
+  "<|generator_expression_py -> if_clause_py|>": 152098,
+  "<|generic_type_py -> identifier_py type_parameter_py |>": 151925,
+  "<|if_clause_py -> if_ter attribute_py |>": 151789,
+  "<|if_clause_py -> if_ter binary_operator_py |>": 151873,
+  "<|if_clause_py -> if_ter boolean_operator_py |>": 151783,
+  "<|if_clause_py -> if_ter call_py |>": 151847,
+  "<|if_clause_py -> if_ter comparison_operator_py |>": 152536,
+  "<|if_clause_py -> if_ter false_py |>": 152240,
+  "<|if_clause_py -> if_ter identifier_py |>": 152537,
+  "<|if_clause_py -> if_ter identifier_ter |>": 151875,
+  "<|if_clause_py -> if_ter list_comprehension_py |>": 152226,
+  "<|if_clause_py -> if_ter not_operator_py |>": 151812,
+  "<|if_clause_py -> if_ter parenthesized_expression_py |>": 152213,
+  "<|if_clause_py -> if_ter subscript_py |>": 152126,
+  "<|if_statement_py -> :_ter|>": 152235,
+  "<|if_statement_py -> End |>": 152128,
+  "<|if_statement_py -> alternative_py|>": 152161,
+  "<|if_statement_py -> condition_py|>": 151788,
+  "<|if_statement_py -> consequence_py|>": 152137,
+  "<|if_statement_py -> if_ter|>": 152508,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|import_from_statement_py -> ,_ter|>": 152195,
+  "<|import_from_statement_py -> End |>": 152258,
+  "<|import_from_statement_py -> from_ter|>": 151854,
+  "<|import_from_statement_py -> import_ter|>": 152169,
+  "<|import_from_statement_py -> module_name_py|>": 152180,
+  "<|import_from_statement_py -> name_py|>": 152288,
+  "<|import_statement_py -> ,_ter|>": 152503,
+  "<|import_statement_py -> End |>": 152481,
+  "<|import_statement_py -> import_ter|>": 152376,
+  "<|import_statement_py -> name_py|>": 152143,
+  "<|is_not_py -> is_ter not_ter |>": 151935,
+  "<|key_py -> attribute_py |>": 152200,
+  "<|key_py -> binary_operator_py |>": 152189,
+  "<|key_py -> boolean_operator_py |>": 152463,
+  "<|key_py -> call_py |>": 152253,
+  "<|key_py -> comparison_operator_py |>": 151691,
+  "<|key_py -> conditional_expression_py |>": 151905,
+  "<|key_py -> dictionary_py |>": 152012,
+  "<|key_py -> false_py |>": 151855,
+  "<|key_py -> float_py |>": 152096,
+  "<|key_py -> identifier_py |>": 151960,
+  "<|key_py -> identifier_ter |>": 152469,
+  "<|key_py -> integer_py |>": 152028,
+  "<|key_py -> lambda_py |>": 152294,
+  "<|key_py -> list_py |>": 151985,
+  "<|key_py -> none_py |>": 151711,
+  "<|key_py -> parenthesized_expression_py |>": 152238,
+  "<|key_py -> set_py |>": 151853,
+  "<|key_py -> string_literal_py |>": 152050,
+  "<|key_py -> subscript_py |>": 151749,
+  "<|key_py -> true_py |>": 152395,
+  "<|key_py -> tuple_py |>": 151839,
+  "<|key_py -> unary_operator_py |>": 152524,
+  "<|keyword_argument_py -> name_py =_ter value_py |>": 152406,
+  "<|lambda_parameters_py -> ,_ter|>": 151844,
+  "<|lambda_parameters_py -> End |>": 152191,
+  "<|lambda_parameters_py -> default_parameter_py|>": 152091,
+  "<|lambda_parameters_py -> identifier_py|>": 151860,
+  "<|lambda_parameters_py -> identifier_ter|>": 151842,
+  "<|lambda_py -> lambda_ter :_ter body_py |>": 151773,
+  "<|lambda_py -> lambda_ter parameters_py :_ter body_py |>": 152432,
+  "<|left_py -> attribute_py |>": 151893,
+  "<|left_py -> binary_operator_py |>": 151907,
+  "<|left_py -> boolean_operator_py |>": 152321,
+  "<|left_py -> call_py |>": 152366,
+  "<|left_py -> comparison_operator_py |>": 151871,
+  "<|left_py -> dictionary_comprehension_py |>": 151667,
+  "<|left_py -> dictionary_py |>": 152337,
+  "<|left_py -> false_py |>": 152302,
+  "<|left_py -> float_py |>": 152233,
+  "<|left_py -> generator_expression_py |>": 151917,
+  "<|left_py -> identifier_py |>": 152029,
+  "<|left_py -> identifier_ter |>": 152224,
+  "<|left_py -> integer_py |>": 151830,
+  "<|left_py -> list_comprehension_py |>": 152297,
+  "<|left_py -> list_py |>": 151782,
+  "<|left_py -> none_py |>": 152174,
+  "<|left_py -> not_operator_py |>": 152439,
+  "<|left_py -> parenthesized_expression_py |>": 152127,
+  "<|left_py -> pattern_list_py |>": 151765,
+  "<|left_py -> set_comprehension_py |>": 152016,
+  "<|left_py -> set_py |>": 151790,
+  "<|left_py -> string_literal_py |>": 152444,
+  "<|left_py -> subscript_py |>": 152345,
+  "<|left_py -> true_py |>": 152057,
+  "<|left_py -> tuple_pattern_py |>": 152199,
+  "<|left_py -> tuple_py |>": 151838,
+  "<|left_py -> type_py |>": 152349,
+  "<|left_py -> unary_operator_py |>": 151810,
+  "<|list_comprehension_py -> End |>": 152391,
+  "<|list_comprehension_py -> [_ter|>": 151845,
+  "<|list_comprehension_py -> ]_ter|>": 151780,
+  "<|list_comprehension_py -> body_py|>": 152056,
+  "<|list_comprehension_py -> for_in_clause_py|>": 151826,
+  "<|list_comprehension_py -> if_clause_py|>": 152528,
+  "<|list_py -> ,_ter|>": 152198,
+  "<|list_py -> End |>": 151991,
+  "<|list_py -> [_ter|>": 151709,
+  "<|list_py -> ]_ter|>": 152414,
+  "<|list_py -> attribute_py|>": 151728,
+  "<|list_py -> binary_operator_py|>": 151872,
+  "<|list_py -> boolean_operator_py|>": 152460,
+  "<|list_py -> call_py|>": 151937,
+  "<|list_py -> comparison_operator_py|>": 152389,
+  "<|list_py -> conditional_expression_py|>": 152353,
+  "<|list_py -> dictionary_comprehension_py|>": 151980,
+  "<|list_py -> dictionary_py|>": 151885,
+  "<|list_py -> false_py|>": 152482,
+  "<|list_py -> float_py|>": 152058,
+  "<|list_py -> generator_expression_py|>": 151863,
+  "<|list_py -> identifier_py|>": 152364,
+  "<|list_py -> identifier_ter|>": 152304,
+  "<|list_py -> integer_py|>": 152458,
+  "<|list_py -> lambda_py|>": 152084,
+  "<|list_py -> list_comprehension_py|>": 151919,
+  "<|list_py -> list_py|>": 151865,
+  "<|list_py -> list_splat_py|>": 151698,
+  "<|list_py -> none_py|>": 151784,
+  "<|list_py -> not_operator_py|>": 151713,
+  "<|list_py -> parenthesized_expression_py|>": 151823,
+  "<|list_py -> set_comprehension_py|>": 152424,
+  "<|list_py -> set_py|>": 152278,
+  "<|list_py -> string_literal_py|>": 151738,
+  "<|list_py -> subscript_py|>": 151752,
+  "<|list_py -> true_py|>": 152008,
+  "<|list_py -> tuple_py|>": 151840,
+  "<|list_py -> unary_operator_py|>": 151896,
+  "<|list_splat_pattern_py -> *_ter attribute_py |>": 151874,
+  "<|list_splat_py -> *_ter attribute_py |>": 152344,
+  "<|list_splat_py -> *_ter binary_operator_py |>": 152285,
+  "<|list_splat_py -> *_ter boolean_operator_py |>": 152401,
+  "<|list_splat_py -> *_ter call_py |>": 152041,
+  "<|list_splat_py -> *_ter identifier_py |>": 152207,
+  "<|list_splat_py -> *_ter subscript_py |>": 152073,
+  "<|module_name_py -> dotted_name_py |>": 151903,
+  "<|module_py -> ;_ter|>": 152211,
+  "<|module_py -> End |>": 152064,
+  "<|module_py -> assert_statement_py|>": 152201,
+  "<|module_py -> break_statement_py|>": 151732,
+  "<|module_py -> class_definition_py|>": 152526,
+  "<|module_py -> delete_statement_py|>": 152329,
+  "<|module_py -> expression_statement_py|>": 152142,
+  "<|module_py -> for_statement_py|>": 151934,
+  "<|module_py -> function_definition_py|>": 151757,
+  "<|module_py -> future_import_statement_py|>": 152529,
+  "<|module_py -> if_statement_py|>": 151787,
+  "<|module_py -> import_from_statement_py|>": 152242,
+  "<|module_py -> import_statement_py|>": 152270,
+  "<|module_py -> pass_statement_py|>": 151772,
+  "<|module_py -> print_statement_py|>": 151764,
+  "<|module_py -> return_statement_py|>": 152343,
+  "<|module_py -> try_statement_py|>": 152330,
+  "<|module_py -> type_alias_statement_py|>": 151792,
+  "<|module_py -> while_statement_py|>": 151834,
+  "<|name_py -> aliased_import_py |>": 151694,
+  "<|name_py -> dotted_name_py |>": 152150,
+  "<|name_py -> identifier_py |>": 151828,
+  "<|name_py -> identifier_ter |>": 152425,
+  "<|none_py -> None_ter |>": 151751,
+  "<|not_in_py -> not_ter in_ter |>": 152472,
+  "<|not_operator_py -> not_ter argument_py |>": 151928,
+  "<|object_py -> attribute_py |>": 152155,
+  "<|object_py -> call_py |>": 152100,
+  "<|object_py -> dictionary_comprehension_py |>": 152116,
+  "<|object_py -> dictionary_py |>": 152237,
+  "<|object_py -> float_py |>": 151888,
+  "<|object_py -> generator_expression_py |>": 152445,
+  "<|object_py -> identifier_py |>": 152190,
+  "<|object_py -> identifier_ter |>": 152265,
+  "<|object_py -> integer_py |>": 152339,
+  "<|object_py -> list_comprehension_py |>": 152160,
+  "<|object_py -> list_py |>": 151781,
+  "<|object_py -> list_splat_py |>": 152119,
+  "<|object_py -> none_py |>": 152024,
+  "<|object_py -> parenthesized_expression_py |>": 152196,
+  "<|object_py -> set_comprehension_py |>": 152418,
+  "<|object_py -> set_py |>": 152477,
+  "<|object_py -> string_literal_py |>": 151705,
+  "<|object_py -> subscript_py |>": 151977,
+  "<|object_py -> tuple_py |>": 152217,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|operator_py -> %=_ter |>": 152159,
+  "<|operator_py -> %_ter |>": 151734,
+  "<|operator_py -> &_ter |>": 152114,
+  "<|operator_py -> **_ter |>": 152139,
+  "<|operator_py -> *=_ter |>": 151849,
+  "<|operator_py -> *_ter |>": 152499,
+  "<|operator_py -> +=_ter |>": 152394,
+  "<|operator_py -> +_ter |>": 152374,
+  "<|operator_py -> -=_ter |>": 152210,
+  "<|operator_py -> -_ter |>": 152080,
+  "<|operator_py -> //=_ter |>": 152289,
+  "<|operator_py -> //_ter |>": 152392,
+  "<|operator_py -> /=_ter |>": 152140,
+  "<|operator_py -> /_ter |>": 152396,
+  "<|operator_py -> <<=_ter |>": 152450,
+  "<|operator_py -> <<_ter |>": 152053,
+  "<|operator_py -> >>=_ter |>": 152336,
+  "<|operator_py -> >>_ter |>": 151683,
+  "<|operator_py -> ^_ter |>": 152141,
+  "<|operator_py -> and_ter |>": 151735,
+  "<|operator_py -> or_ter |>": 152328,
+  "<|operator_py -> |=_ter |>": 152078,
+  "<|operator_py -> |_ter |>": 151726,
+  "<|operator_py -> ~_ter |>": 151864,
+  "<|operators_py -> !=_ter |>": 151880,
+  "<|operators_py -> <=_ter |>": 151676,
+  "<|operators_py -> <_ter |>": 152468,
+  "<|operators_py -> ==_ter |>": 152399,
+  "<|operators_py -> >=_ter |>": 152167,
+  "<|operators_py -> >_ter |>": 152434,
+  "<|operators_py -> in_ter |>": 152377,
+  "<|operators_py -> is_not_py |>": 151898,
+  "<|operators_py -> is_ter |>": 152117,
+  "<|operators_py -> not_in_py |>": 152531,
+  "<|pair_py -> key_py :_ter value_py |>": 151681,
+  "<|parameters_py -> (_ter|>": 151731,
+  "<|parameters_py -> )_ter|>": 151714,
+  "<|parameters_py -> ,_ter|>": 152040,
+  "<|parameters_py -> End |>": 152370,
+  "<|parameters_py -> default_parameter_py|>": 152197,
+  "<|parameters_py -> identifier_py|>": 152308,
+  "<|parameters_py -> identifier_ter|>": 152070,
+  "<|parameters_py -> lambda_parameters_py|>": 152520,
+  "<|parameters_py -> parameters_py|>": 152082,
+  "<|parameters_py -> typed_parameter_py|>": 151998,
+  "<|parenthesized_expression_py -> (_ter attribute_py )_ter |>": 151665,
+  "<|parenthesized_expression_py -> (_ter binary_operator_py )_ter |>": 152185,
+  "<|parenthesized_expression_py -> (_ter boolean_operator_py )_ter |>": 151716,
+  "<|parenthesized_expression_py -> (_ter call_py )_ter |>": 152397,
+  "<|parenthesized_expression_py -> (_ter comparison_operator_py )_ter |>": 152136,
+  "<|parenthesized_expression_py -> (_ter conditional_expression_py )_ter |>": 151887,
+  "<|parenthesized_expression_py -> (_ter dictionary_comprehension_py )_ter |>": 151715,
+  "<|parenthesized_expression_py -> (_ter dictionary_py )_ter |>": 152234,
+  "<|parenthesized_expression_py -> (_ter false_py )_ter |>": 151677,
+  "<|parenthesized_expression_py -> (_ter float_py )_ter |>": 152493,
+  "<|parenthesized_expression_py -> (_ter identifier_py )_ter |>": 151861,
+  "<|parenthesized_expression_py -> (_ter integer_py )_ter |>": 152332,
+  "<|parenthesized_expression_py -> (_ter lambda_py )_ter |>": 152398,
+  "<|parenthesized_expression_py -> (_ter list_comprehension_py )_ter |>": 151833,
+  "<|parenthesized_expression_py -> (_ter list_py )_ter |>": 152193,
+  "<|parenthesized_expression_py -> (_ter not_operator_py )_ter |>": 152507,
+  "<|parenthesized_expression_py -> (_ter parenthesized_expression_py )_ter |>": 151981,
+  "<|parenthesized_expression_py -> (_ter set_py )_ter |>": 152266,
+  "<|parenthesized_expression_py -> (_ter string_literal_py )_ter |>": 151967,
+  "<|parenthesized_expression_py -> (_ter subscript_py )_ter |>": 151774,
+  "<|parenthesized_expression_py -> (_ter true_py )_ter |>": 152305,
+  "<|parenthesized_expression_py -> (_ter tuple_py )_ter |>": 152168,
+  "<|parenthesized_expression_py -> (_ter unary_operator_py )_ter |>": 151976,
+  "<|parenthesized_expression_py -> (_ter yield_py )_ter |>": 151890,
+  "<|pass_statement_py -> pass_ter |>": 152152,
+  "<|pattern_list_py -> ,_ter|>": 152214,
+  "<|pattern_list_py -> End |>": 152042,
+  "<|pattern_list_py -> attribute_py|>": 152262,
+  "<|pattern_list_py -> identifier_py|>": 152031,
+  "<|pattern_list_py -> identifier_ter|>": 152412,
+  "<|pattern_list_py -> list_splat_pattern_py|>": 151990,
+  "<|pattern_list_py -> subscript_py|>": 151776,
+  "<|pattern_list_py -> tuple_pattern_py|>": 152079,
+  "<|print_statement_py -> ,_ter|>": 151727,
+  "<|print_statement_py -> End |>": 152173,
+  "<|print_statement_py -> argument_py|>": 151929,
+  "<|print_statement_py -> chevron_py|>": 151753,
+  "<|print_statement_py -> print_ter|>": 152243,
+  "<|python -> module_py |>": 152274,
+  "<|python -> module_ter |>": 151796,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|raise_statement_py -> raise_ter |>": 151723,
+  "<|repo_name|>": 151663,
+  "<|return_statement_py -> return_ter attribute_py |>": 152188,
+  "<|return_statement_py -> return_ter binary_operator_py |>": 151754,
+  "<|return_statement_py -> return_ter boolean_operator_py |>": 152295,
+  "<|return_statement_py -> return_ter call_py |>": 152429,
+  "<|return_statement_py -> return_ter comparison_operator_py |>": 152375,
+  "<|return_statement_py -> return_ter conditional_expression_py |>": 152473,
+  "<|return_statement_py -> return_ter dictionary_comprehension_py |>": 151997,
+  "<|return_statement_py -> return_ter dictionary_py |>": 151825,
+  "<|return_statement_py -> return_ter expression_list_py |>": 151671,
+  "<|return_statement_py -> return_ter false_py |>": 152522,
+  "<|return_statement_py -> return_ter float_py |>": 152292,
+  "<|return_statement_py -> return_ter generator_expression_py |>": 151918,
+  "<|return_statement_py -> return_ter identifier_py |>": 152062,
+  "<|return_statement_py -> return_ter identifier_ter |>": 152410,
+  "<|return_statement_py -> return_ter integer_py |>": 151892,
+  "<|return_statement_py -> return_ter lambda_py |>": 152324,
+  "<|return_statement_py -> return_ter list_comprehension_py |>": 152539,
+  "<|return_statement_py -> return_ter list_py |>": 152025,
+  "<|return_statement_py -> return_ter list_splat_py |>": 151945,
+  "<|return_statement_py -> return_ter none_py |>": 152437,
+  "<|return_statement_py -> return_ter not_operator_py |>": 152015,
+  "<|return_statement_py -> return_ter parenthesized_expression_py |>": 152246,
+  "<|return_statement_py -> return_ter set_comprehension_py |>": 151802,
+  "<|return_statement_py -> return_ter set_py |>": 152513,
+  "<|return_statement_py -> return_ter string_literal_py |>": 151954,
+  "<|return_statement_py -> return_ter subscript_py |>": 152504,
+  "<|return_statement_py -> return_ter true_py |>": 151769,
+  "<|return_statement_py -> return_ter tuple_py |>": 151921,
+  "<|return_statement_py -> return_ter unary_operator_py |>": 152149,
+  "<|return_statement_py -> return_ter |>": 152093,
+  "<|return_type_py -> type_py |>": 152441,
+  "<|right_py -> assignment_py |>": 152075,
+  "<|right_py -> attribute_py |>": 151743,
+  "<|right_py -> binary_operator_py |>": 152403,
+  "<|right_py -> boolean_operator_py |>": 152081,
+  "<|right_py -> call_py |>": 151680,
+  "<|right_py -> comparison_operator_py |>": 151870,
+  "<|right_py -> conditional_expression_py |>": 152451,
+  "<|right_py -> dictionary_comprehension_py |>": 152402,
+  "<|right_py -> dictionary_py |>": 152313,
+  "<|right_py -> expression_list_py |>": 152055,
+  "<|right_py -> false_py |>": 151696,
+  "<|right_py -> float_py |>": 151942,
+  "<|right_py -> generator_expression_py |>": 151688,
+  "<|right_py -> identifier_py |>": 151879,
+  "<|right_py -> identifier_ter |>": 151814,
+  "<|right_py -> integer_py |>": 151785,
+  "<|right_py -> lambda_py |>": 151766,
+  "<|right_py -> list_comprehension_py |>": 152273,
+  "<|right_py -> list_py |>": 152129,
+  "<|right_py -> list_splat_py |>": 151904,
+  "<|right_py -> none_py |>": 151856,
+  "<|right_py -> not_operator_py |>": 152358,
+  "<|right_py -> parenthesized_expression_py |>": 152144,
+  "<|right_py -> pattern_list_py |>": 152092,
+  "<|right_py -> set_comprehension_py |>": 151911,
+  "<|right_py -> set_py |>": 151682,
+  "<|right_py -> string_literal_py |>": 152094,
+  "<|right_py -> subscript_py |>": 152138,
+  "<|right_py -> true_py |>": 152342,
+  "<|right_py -> tuple_py |>": 151926,
+  "<|right_py -> type_py |>": 152216,
+  "<|right_py -> unary_operator_py |>": 151684,
+  "<|right_py -> yield_py |>": 152293,
+  "<|set_comprehension_py -> End |>": 151809,
+  "<|set_comprehension_py -> body_py|>": 151988,
+  "<|set_comprehension_py -> for_in_clause_py|>": 152134,
+  "<|set_comprehension_py -> if_clause_py|>": 152380,
+  "<|set_comprehension_py -> {_ter|>": 151704,
+  "<|set_comprehension_py -> }_ter|>": 152440,
+  "<|set_py -> ,_ter|>": 151811,
+  "<|set_py -> End |>": 152505,
+  "<|set_py -> attribute_py|>": 151841,
+  "<|set_py -> binary_operator_py|>": 152171,
+  "<|set_py -> call_py|>": 152077,
+  "<|set_py -> comparison_operator_py|>": 151943,
+  "<|set_py -> conditional_expression_py|>": 152338,
+  "<|set_py -> dictionary_py|>": 152490,
+  "<|set_py -> false_py|>": 152014,
+  "<|set_py -> float_py|>": 151815,
+  "<|set_py -> identifier_py|>": 151672,
+  "<|set_py -> integer_py|>": 152225,
+  "<|set_py -> lambda_py|>": 152323,
+  "<|set_py -> list_py|>": 152404,
+  "<|set_py -> list_splat_py|>": 151720,
+  "<|set_py -> none_py|>": 151836,
+  "<|set_py -> parenthesized_expression_py|>": 152130,
+  "<|set_py -> set_py|>": 152051,
+  "<|set_py -> string_literal_py|>": 152484,
+  "<|set_py -> subscript_py|>": 151966,
+  "<|set_py -> true_py|>": 152511,
+  "<|set_py -> tuple_py|>": 152365,
+  "<|set_py -> unary_operator_py|>": 151756,
+  "<|set_py -> {_ter|>": 152427,
+  "<|set_py -> }_ter|>": 151927,
+  "<|slice_py -> :_ter|>": 152494,
+  "<|slice_py -> End |>": 152032,
+  "<|slice_py -> attribute_py|>": 152303,
+  "<|slice_py -> binary_operator_py|>": 152497,
+  "<|slice_py -> boolean_operator_py|>": 152301,
+  "<|slice_py -> call_py|>": 152227,
+  "<|slice_py -> conditional_expression_py|>": 152124,
+  "<|slice_py -> identifier_py|>": 152215,
+  "<|slice_py -> integer_py|>": 152252,
+  "<|slice_py -> none_py|>": 152006,
+  "<|slice_py -> parenthesized_expression_py|>": 152209,
+  "<|slice_py -> string_literal_py|>": 152118,
+  "<|slice_py -> subscript_py|>": 152299,
+  "<|slice_py -> unary_operator_py|>": 152489,
+  "<|start -> python|>": 152298,
+  "<|string_literal_py -> End|>": 152476,
+  "<|subscript_py -> End |>": 152020,
+  "<|subscript_py -> [_ter|>": 151946,
+  "<|subscript_py -> ]_ter|>": 151940,
+  "<|subscript_py -> attribute_py|>": 152416,
+  "<|subscript_py -> binary_operator_py|>": 152415,
+  "<|subscript_py -> boolean_operator_py|>": 152352,
+  "<|subscript_py -> call_py|>": 152101,
+  "<|subscript_py -> comparison_operator_py|>": 152218,
+  "<|subscript_py -> conditional_expression_py|>": 151867,
+  "<|subscript_py -> dictionary_py|>": 152005,
+  "<|subscript_py -> false_py|>": 151702,
+  "<|subscript_py -> float_py|>": 151881,
+  "<|subscript_py -> generator_expression_py|>": 152108,
+  "<|subscript_py -> identifier_py|>": 152148,
+  "<|subscript_py -> identifier_ter|>": 151978,
+  "<|subscript_py -> integer_py|>": 152272,
+  "<|subscript_py -> lambda_py|>": 152506,
+  "<|subscript_py -> list_comprehension_py|>": 152204,
+  "<|subscript_py -> list_py|>": 152001,
+  "<|subscript_py -> none_py|>": 151813,
+  "<|subscript_py -> not_operator_py|>": 152362,
+  "<|subscript_py -> parenthesized_expression_py|>": 152316,
+  "<|subscript_py -> set_py|>": 152386,
+  "<|subscript_py -> slice_py|>": 151848,
+  "<|subscript_py -> string_literal_py|>": 152220,
+  "<|subscript_py -> subscript_py|>": 151894,
+  "<|subscript_py -> true_py|>": 151708,
+  "<|subscript_py -> tuple_py|>": 152419,
+  "<|subscript_py -> unary_operator_py|>": 151987,
+  "<|subscript_py -> value_py|>": 151725,
+  "<|superclasses_py -> argument_list_py |>": 151899,
+  "<|true_py -> True_ter |>": 151747,
+  "<|try_statement_py -> :_ter|>": 151706,
+  "<|try_statement_py -> End |>": 151982,
+  "<|try_statement_py -> body_py|>": 152383,
+  "<|try_statement_py -> except_clause_py|>": 152500,
+  "<|try_statement_py -> try_ter|>": 151850,
+  "<|tuple_pattern_py -> (_ter|>": 152121,
+  "<|tuple_pattern_py -> )_ter|>": 152277,
+  "<|tuple_pattern_py -> ,_ter|>": 151920,
+  "<|tuple_pattern_py -> End |>": 152267,
+  "<|tuple_pattern_py -> attribute_py|>": 152535,
+  "<|tuple_pattern_py -> identifier_py|>": 152371,
+  "<|tuple_pattern_py -> identifier_ter|>": 152541,
+  "<|tuple_pattern_py -> subscript_py|>": 152417,
+  "<|tuple_pattern_py -> tuple_pattern_py|>": 152249,
+  "<|tuple_py -> (_ter|>": 152254,
+  "<|tuple_py -> )_ter|>": 152076,
+  "<|tuple_py -> ,_ter|>": 151829,
+  "<|tuple_py -> End |>": 151909,
+  "<|tuple_py -> attribute_py|>": 152132,
+  "<|tuple_py -> binary_operator_py|>": 152501,
+  "<|tuple_py -> boolean_operator_py|>": 152360,
+  "<|tuple_py -> call_py|>": 152331,
+  "<|tuple_py -> comparison_operator_py|>": 152013,
+  "<|tuple_py -> conditional_expression_py|>": 151689,
+  "<|tuple_py -> dictionary_comprehension_py|>": 151687,
+  "<|tuple_py -> dictionary_py|>": 152462,
+  "<|tuple_py -> false_py|>": 151748,
+  "<|tuple_py -> float_py|>": 151956,
+  "<|tuple_py -> generator_expression_py|>": 152443,
+  "<|tuple_py -> identifier_py|>": 152381,
+  "<|tuple_py -> identifier_ter|>": 152021,
+  "<|tuple_py -> integer_py|>": 152157,
+  "<|tuple_py -> lambda_py|>": 151761,
+  "<|tuple_py -> list_comprehension_py|>": 152186,
+  "<|tuple_py -> list_py|>": 152518,
+  "<|tuple_py -> list_splat_py|>": 152085,
+  "<|tuple_py -> none_py|>": 152290,
+  "<|tuple_py -> not_operator_py|>": 151819,
+  "<|tuple_py -> parenthesized_expression_py|>": 152368,
+  "<|tuple_py -> set_comprehension_py|>": 152135,
+  "<|tuple_py -> set_py|>": 152071,
+  "<|tuple_py -> string_literal_py|>": 152459,
+  "<|tuple_py -> subscript_py|>": 152334,
+  "<|tuple_py -> true_py|>": 151832,
+  "<|tuple_py -> tuple_py|>": 152264,
+  "<|tuple_py -> unary_operator_py|>": 152320,
+  "<|type_alias_statement_py -> type_ter left_py =_ter right_py |>": 151701,
+  "<|type_parameter_py -> ,_ter|>": 152170,
+  "<|type_parameter_py -> End |>": 151979,
+  "<|type_parameter_py -> [_ter|>": 151768,
+  "<|type_parameter_py -> ]_ter|>": 151964,
+  "<|type_parameter_py -> type_py|>": 152296,
+  "<|type_py -> attribute_py |>": 151924,
+  "<|type_py -> binary_operator_py |>": 152037,
+  "<|type_py -> boolean_operator_py |>": 151957,
+  "<|type_py -> call_py |>": 152257,
+  "<|type_py -> dictionary_py |>": 152241,
+  "<|type_py -> false_py |>": 151710,
+  "<|type_py -> generic_type_py |>": 152177,
+  "<|type_py -> identifier_py |>": 151923,
+  "<|type_py -> integer_py |>": 152534,
+  "<|type_py -> list_py |>": 151669,
+  "<|type_py -> none_py |>": 151700,
+  "<|type_py -> parenthesized_expression_py |>": 152467,
+  "<|type_py -> set_py |>": 152067,
+  "<|type_py -> string_literal_py |>": 152245,
+  "<|type_py -> subscript_py |>": 151807,
+  "<|type_py -> true_py |>": 152131,
+  "<|type_py -> tuple_py |>": 152103,
+  "<|type_py -> type_py |>": 151666,
+  "<|type_py -> unary_operator_py |>": 151969,
+  "<|typed_parameter_py -> identifier_py :_ter type_py |>": 152231,
+  "<|typed_parameter_py -> identifier_ter :_ter type_py |>": 152435,
+  "<|unary_operator_py -> operator_py argument_py |>": 151767,
+  "<|value_py -> attribute_py |>": 152263,
+  "<|value_py -> binary_operator_py |>": 152194,
+  "<|value_py -> boolean_operator_py |>": 151798,
+  "<|value_py -> call_py |>": 151851,
+  "<|value_py -> comparison_operator_py |>": 151984,
+  "<|value_py -> conditional_expression_py |>": 151685,
+  "<|value_py -> dictionary_comprehension_py |>": 152017,
+  "<|value_py -> dictionary_py |>": 152251,
+  "<|value_py -> false_py |>": 152525,
+  "<|value_py -> float_py |>": 152065,
+  "<|value_py -> generator_expression_py |>": 152166,
+  "<|value_py -> identifier_py |>": 152110,
+  "<|value_py -> identifier_ter |>": 151944,
+  "<|value_py -> integer_py |>": 151883,
+  "<|value_py -> lambda_py |>": 152244,
+  "<|value_py -> list_comprehension_py |>": 152483,
+  "<|value_py -> list_py |>": 152540,
+  "<|value_py -> list_splat_py |>": 151822,
+  "<|value_py -> none_py |>": 151808,
+  "<|value_py -> not_operator_py |>": 152538,
+  "<|value_py -> parenthesized_expression_py |>": 151843,
+  "<|value_py -> set_comprehension_py |>": 152004,
+  "<|value_py -> set_py |>": 152433,
+  "<|value_py -> string_literal_py |>": 152413,
+  "<|value_py -> subscript_py |>": 151707,
+  "<|value_py -> true_py |>": 151786,
+  "<|value_py -> tuple_py |>": 152202,
+  "<|value_py -> unary_operator_py |>": 152442,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652,
+  "<|while_statement_py -> while_ter condition_py :_ter body_py |>": 152519,
+  "<|yield_py -> yield_ter attribute_py |>": 152046,
+  "<|yield_py -> yield_ter binary_operator_py |>": 151932,
+  "<|yield_py -> yield_ter call_py |>": 152367,
+  "<|yield_py -> yield_ter comparison_operator_py |>": 151939,
+  "<|yield_py -> yield_ter conditional_expression_py |>": 151993,
+  "<|yield_py -> yield_ter dictionary_comprehension_py |>": 152010,
+  "<|yield_py -> yield_ter dictionary_py |>": 151877,
+  "<|yield_py -> yield_ter expression_list_py |>": 152409,
+  "<|yield_py -> yield_ter false_py |>": 151778,
+  "<|yield_py -> yield_ter float_py |>": 151744,
+  "<|yield_py -> yield_ter from_ter attribute_py |>": 152018,
+  "<|yield_py -> yield_ter from_ter call_py |>": 151827,
+  "<|yield_py -> yield_ter from_ter conditional_expression_py |>": 151869,
+  "<|yield_py -> yield_ter from_ter generator_expression_py |>": 151741,
+  "<|yield_py -> yield_ter from_ter identifier_py |>": 151771,
+  "<|yield_py -> yield_ter from_ter list_py |>": 152123,
+  "<|yield_py -> yield_ter from_ter subscript_py |>": 152532,
+  "<|yield_py -> yield_ter from_ter tuple_py |>": 151974,
+  "<|yield_py -> yield_ter identifier_py |>": 152382,
+  "<|yield_py -> yield_ter integer_py |>": 151675,
+  "<|yield_py -> yield_ter lambda_py |>": 151718,
+  "<|yield_py -> yield_ter list_comprehension_py |>": 152378,
+  "<|yield_py -> yield_ter list_py |>": 152455,
+  "<|yield_py -> yield_ter none_py |>": 151948,
+  "<|yield_py -> yield_ter parenthesized_expression_py |>": 152474,
+  "<|yield_py -> yield_ter string_literal_py |>": 152172,
+  "<|yield_py -> yield_ter subscript_py |>": 152283,
+  "<|yield_py -> yield_ter true_py |>": 151914,
+  "<|yield_py -> yield_ter tuple_py |>": 152400,
+  "<|yield_py -> yield_ter unary_operator_py |>": 151690,
+  "<|yield_py -> yield_ter |>": 152105
+}

config.json

@@ -0,0 +1,30 @@
+{
+  "_name_or_path": "/share/zhangzhao/models/qwen2.5-1.5b-grammar",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 131072,
+  "max_window_layers": 28,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 131072,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": false,
+  "use_mrope": false,
+  "use_sliding_window": false,
+  "vocab_size": 152542
+}

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.49.0"
+}

global_step34799/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b16ca33fc1c328ba4f18616efbd3783bf352ebe7bb439e552972e4818fbc0221
+size 2316974597

global_step34799/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a13afb7c08d34a8a9774a4d853880658cdcbd994d147cf614536a2d6c740a53
+size 2316976389

global_step34799/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c29e6549e6176e8de7d30a61979b35c31922918818056f1ac53b7f863711fb05
+size 2316976645

global_step34799/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:200f6ef068a8e126944bad540d92c0e414263875dca701f001c338989548412d
+size 2316976645

global_step34799/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2d6a4558aa889b025fd749c022bcd43d4d58fde9998c3fbf18e0281d2f61f5c
+size 2316976709

global_step34799/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:855552ad5df40c2afea18dc8d05f547d075a6a979a1ea1a604bb4159f7d714a0
+size 2316976645

global_step34799/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c4ab5b1a54fbce321b6aee4d3e8964af54eb785c203ee89bb3f51c4bf7bd8e7
+size 2316976709

global_step34799/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3299d4fb1d30616cd4d2725251e3ebac9cd92baf82452f7decc311eed382c8f4
+size 2316976645

global_step34799/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e041639319eafda30dcefe57ae5545cc47fd7ceb272b8b8463bdbcebc04376a
+size 3089382501

latest

@@ -0,0 +1 @@
+global_step34799

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bcd04c7c1da432d5c1c7d291ec48e28cb1cefc69f65d33fe3e97968e976e6e7
+size 3557938016

rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74c8d337722463e03302f4febf685dad8b5696faf483eba44fc08dad412ada63
+size 16389

rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53c4237c1614c4c633a43025a0090a990233816338927c9c955846324d5b2a50
+size 16389

rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ef41f6e29cc1a22a05075b00f43810e0c57a9da1e5acbaeb5dbc75ba17687c8
+size 16389

rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:230b943449a21bb79084e2e41f16597020eec2c876a22f1a50b7f9e90aeb9a03
+size 16389

rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9fb5842a00fbaaf416fd46e2a48f3cc0ebb987e5f71c6539a053019b2579d2e
+size 16389

rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:beba699c5fc429b85822cd284c107538cf35942d1dfee2eed2811b645857b2c2
+size 16389

rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c0409921d93b5a657666afceacaf68ace4bec5b421ced13ace851fc935778c6
+size 16389

rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1db8beb7baf2946f5820ec75d95bd275110ea83cd4d2c10904804078dddf05a2
+size 16389

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1199af05b688039c2c15282e083daca1652c50fba158ac89689cd0fc819bb3c
+size 1465

special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:186f6baba553b452869bf93d7cc258f82498026c81af406284ac20781c6b8173
+size 11609235

trainer_state.json

@@ -0,0 +1,1251 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.077741172308577,
+  "eval_steps": 500,
+  "global_step": 34800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.017688548875671613,
+      "grad_norm": 4.9453349113464355,
+      "learning_rate": 1.768972227136034e-06,
+      "loss": 6.7403,
+      "step": 200
+    },
+    {
+      "epoch": 0.03537709775134323,
+      "grad_norm": 8.81293773651123,
+      "learning_rate": 3.537944454272068e-06,
+      "loss": 4.3531,
+      "step": 400
+    },
+    {
+      "epoch": 0.05306564662701484,
+      "grad_norm": 5.486693382263184,
+      "learning_rate": 5.306916681408102e-06,
+      "loss": 2.2861,
+      "step": 600
+    },
+    {
+      "epoch": 0.07075419550268645,
+      "grad_norm": 2.244082450866699,
+      "learning_rate": 7.075888908544136e-06,
+      "loss": 1.162,
+      "step": 800
+    },
+    {
+      "epoch": 0.08844274437835806,
+      "grad_norm": 0.6459134221076965,
+      "learning_rate": 8.84486113568017e-06,
+      "loss": 0.3574,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10613129325402967,
+      "grad_norm": 0.42250216007232666,
+      "learning_rate": 1.0613833362816204e-05,
+      "loss": 0.22,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12381984212970129,
+      "grad_norm": 0.4047611951828003,
+      "learning_rate": 1.2382805589952239e-05,
+      "loss": 0.1967,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1415083910053729,
+      "grad_norm": 0.34184110164642334,
+      "learning_rate": 1.4151777817088272e-05,
+      "loss": 0.1853,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1591969398810445,
+      "grad_norm": 0.32166364789009094,
+      "learning_rate": 1.5920750044224307e-05,
+      "loss": 0.1795,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1768854887567161,
+      "grad_norm": 0.30309006571769714,
+      "learning_rate": 1.768972227136034e-05,
+      "loss": 0.1777,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19457403763238773,
+      "grad_norm": 0.28535833954811096,
+      "learning_rate": 1.9458694498496373e-05,
+      "loss": 0.1728,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21226258650805935,
+      "grad_norm": 0.2785949409008026,
+      "learning_rate": 2.1227666725632408e-05,
+      "loss": 0.1707,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22995113538373096,
+      "grad_norm": 0.28039732575416565,
+      "learning_rate": 2.2996638952768443e-05,
+      "loss": 0.1682,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24763968425940258,
+      "grad_norm": 0.2688646614551544,
+      "learning_rate": 2.4765611179904478e-05,
+      "loss": 0.1663,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2653282331350742,
+      "grad_norm": 0.2585384249687195,
+      "learning_rate": 2.653458340704051e-05,
+      "loss": 0.1672,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2830167820107458,
+      "grad_norm": 0.2583986222743988,
+      "learning_rate": 2.8303555634176544e-05,
+      "loss": 0.1646,
+      "step": 3200
+    },
+    {
+      "epoch": 0.30070533088641743,
+      "grad_norm": 0.23997661471366882,
+      "learning_rate": 3.007252786131258e-05,
+      "loss": 0.1647,
+      "step": 3400
+    },
+    {
+      "epoch": 0.318393879762089,
+      "grad_norm": 0.2527500092983246,
+      "learning_rate": 3.1841500088448614e-05,
+      "loss": 0.1629,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3360824286377606,
+      "grad_norm": 0.23295536637306213,
+      "learning_rate": 3.3610472315584645e-05,
+      "loss": 0.1615,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3537709775134322,
+      "grad_norm": 0.24165655672550201,
+      "learning_rate": 3.537944454272068e-05,
+      "loss": 0.1626,
+      "step": 4000
+    },
+    {
+      "epoch": 0.37145952638910384,
+      "grad_norm": 0.22286227345466614,
+      "learning_rate": 3.7148416769856715e-05,
+      "loss": 0.1611,
+      "step": 4200
+    },
+    {
+      "epoch": 0.38914807526477546,
+      "grad_norm": 0.23912444710731506,
+      "learning_rate": 3.8917388996992746e-05,
+      "loss": 0.1639,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4068366241404471,
+      "grad_norm": 0.22330383956432343,
+      "learning_rate": 4.0686361224128784e-05,
+      "loss": 0.1613,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4245251730161187,
+      "grad_norm": 0.2139132171869278,
+      "learning_rate": 4.2455333451264816e-05,
+      "loss": 0.1601,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4422137218917903,
+      "grad_norm": 0.19250735640525818,
+      "learning_rate": 4.4224305678400854e-05,
+      "loss": 0.1595,
+      "step": 5000
+    },
+    {
+      "epoch": 0.45990227076746193,
+      "grad_norm": 0.19502046704292297,
+      "learning_rate": 4.5993277905536885e-05,
+      "loss": 0.159,
+      "step": 5200
+    },
+    {
+      "epoch": 0.47759081964313355,
+      "grad_norm": 0.19040720164775848,
+      "learning_rate": 4.776225013267292e-05,
+      "loss": 0.1578,
+      "step": 5400
+    },
+    {
+      "epoch": 0.49527936851880516,
+      "grad_norm": 0.20574906468391418,
+      "learning_rate": 4.9531222359808955e-05,
+      "loss": 0.1603,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5129679173944768,
+      "grad_norm": 0.19954811036586761,
+      "learning_rate": 4.999976891136569e-05,
+      "loss": 0.1582,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5306564662701484,
+      "grad_norm": 0.18808187544345856,
+      "learning_rate": 4.999871234414489e-05,
+      "loss": 0.1589,
+      "step": 6000
+    },
+    {
+      "epoch": 0.54834501514582,
+      "grad_norm": 0.17781756818294525,
+      "learning_rate": 4.999680029474971e-05,
+      "loss": 0.158,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5660335640214916,
+      "grad_norm": 0.18824966251850128,
+      "learning_rate": 4.999403282861248e-05,
+      "loss": 0.1572,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5837221128971632,
+      "grad_norm": 0.20633359253406525,
+      "learning_rate": 4.999041004043882e-05,
+      "loss": 0.1572,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6014106617728349,
+      "grad_norm": 0.17536261677742004,
+      "learning_rate": 4.998593205420432e-05,
+      "loss": 0.1565,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6190992106485064,
+      "grad_norm": 0.1713794469833374,
+      "learning_rate": 4.998059902315038e-05,
+      "loss": 0.1554,
+      "step": 7000
+    },
+    {
+      "epoch": 0.636787759524178,
+      "grad_norm": 0.16785737872123718,
+      "learning_rate": 4.997441112977891e-05,
+      "loss": 0.1526,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6544763083998496,
+      "grad_norm": 0.18514026701450348,
+      "learning_rate": 4.996736858584613e-05,
+      "loss": 0.153,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6721648572755212,
+      "grad_norm": 0.17102603614330292,
+      "learning_rate": 4.995947163235527e-05,
+      "loss": 0.1517,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6898534061511928,
+      "grad_norm": 0.16847601532936096,
+      "learning_rate": 4.9950720539548384e-05,
+      "loss": 0.1519,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7075419550268645,
+      "grad_norm": 0.16419926285743713,
+      "learning_rate": 4.9941115606897036e-05,
+      "loss": 0.1507,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7252305039025361,
+      "grad_norm": 0.17619839310646057,
+      "learning_rate": 4.9930657163092123e-05,
+      "loss": 0.15,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7429190527782077,
+      "grad_norm": 0.16663217544555664,
+      "learning_rate": 4.991934556603254e-05,
+      "loss": 0.1502,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7606076016538793,
+      "grad_norm": 0.18116818368434906,
+      "learning_rate": 4.990718120281304e-05,
+      "loss": 0.1511,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7782961505295509,
+      "grad_norm": 0.1595318764448166,
+      "learning_rate": 4.989416448971088e-05,
+      "loss": 0.1502,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7959846994052225,
+      "grad_norm": 0.15985798835754395,
+      "learning_rate": 4.988029587217165e-05,
+      "loss": 0.1513,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8136732482808942,
+      "grad_norm": 0.17328430712223053,
+      "learning_rate": 4.9865575824794e-05,
+      "loss": 0.1491,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8313617971565658,
+      "grad_norm": 0.1536557823419571,
+      "learning_rate": 4.98500048513134e-05,
+      "loss": 0.1474,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8490503460322374,
+      "grad_norm": 0.15362966060638428,
+      "learning_rate": 4.983358348458491e-05,
+      "loss": 0.1458,
+      "step": 9600
+    },
+    {
+      "epoch": 0.866738894907909,
+      "grad_norm": 0.16235561668872833,
+      "learning_rate": 4.9816312286564926e-05,
+      "loss": 0.1454,
+      "step": 9800
+    },
+    {
+      "epoch": 0.8844274437835806,
+      "grad_norm": 0.6078733801841736,
+      "learning_rate": 4.979819184829197e-05,
+      "loss": 0.148,
+      "step": 10000
+    },
+    {
+      "epoch": 0.9021159926592522,
+      "grad_norm": 0.15071770548820496,
+      "learning_rate": 4.9779222789866476e-05,
+      "loss": 0.1461,
+      "step": 10200
+    },
+    {
+      "epoch": 0.9198045415349239,
+      "grad_norm": 0.15512414276599884,
+      "learning_rate": 4.9759405760429524e-05,
+      "loss": 0.1455,
+      "step": 10400
+    },
+    {
+      "epoch": 0.9374930904105955,
+      "grad_norm": 0.15411154925823212,
+      "learning_rate": 4.9738741438140644e-05,
+      "loss": 0.1437,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9551816392862671,
+      "grad_norm": 0.15488554537296295,
+      "learning_rate": 4.9717230530154657e-05,
+      "loss": 0.1445,
+      "step": 10800
+    },
+    {
+      "epoch": 0.9728701881619387,
+      "grad_norm": 0.14975008368492126,
+      "learning_rate": 4.9694873772597396e-05,
+      "loss": 0.1437,
+      "step": 11000
+    },
+    {
+      "epoch": 0.9905587370376103,
+      "grad_norm": 0.15676584839820862,
+      "learning_rate": 4.967167193054058e-05,
+      "loss": 0.1432,
+      "step": 11200
+    },
+    {
+      "epoch": 1.0082251752271874,
+      "grad_norm": 0.15392336249351501,
+      "learning_rate": 4.964762579797558e-05,
+      "loss": 0.1355,
+      "step": 11400
+    },
+    {
+      "epoch": 1.0259137241028589,
+      "grad_norm": 0.1635395586490631,
+      "learning_rate": 4.962273619778632e-05,
+      "loss": 0.1268,
+      "step": 11600
+    },
+    {
+      "epoch": 1.0436022729785306,
+      "grad_norm": 0.15909574925899506,
+      "learning_rate": 4.959700398172101e-05,
+      "loss": 0.1263,
+      "step": 11800
+    },
+    {
+      "epoch": 1.061290821854202,
+      "grad_norm": 0.16021914780139923,
+      "learning_rate": 4.957043003036311e-05,
+      "loss": 0.1264,
+      "step": 12000
+    },
+    {
+      "epoch": 1.0789793707298738,
+      "grad_norm": 0.163675919175148,
+      "learning_rate": 4.954301525310113e-05,
+      "loss": 0.1262,
+      "step": 12200
+    },
+    {
+      "epoch": 1.0966679196055453,
+      "grad_norm": 0.15520550310611725,
+      "learning_rate": 4.951476058809751e-05,
+      "loss": 0.1275,
+      "step": 12400
+    },
+    {
+      "epoch": 1.114356468481217,
+      "grad_norm": 0.1602706015110016,
+      "learning_rate": 4.948566700225654e-05,
+      "loss": 0.1268,
+      "step": 12600
+    },
+    {
+      "epoch": 1.1320450173568886,
+      "grad_norm": 0.15892288088798523,
+      "learning_rate": 4.945573549119128e-05,
+      "loss": 0.1269,
+      "step": 12800
+    },
+    {
+      "epoch": 1.1497335662325603,
+      "grad_norm": 0.16024959087371826,
+      "learning_rate": 4.9424967079189434e-05,
+      "loss": 0.1265,
+      "step": 13000
+    },
+    {
+      "epoch": 1.1674221151082318,
+      "grad_norm": 0.15324904024600983,
+      "learning_rate": 4.939336281917837e-05,
+      "loss": 0.1265,
+      "step": 13200
+    },
+    {
+      "epoch": 1.1851106639839033,
+      "grad_norm": 0.15968984365463257,
+      "learning_rate": 4.936092379268902e-05,
+      "loss": 0.1269,
+      "step": 13400
+    },
+    {
+      "epoch": 1.202799212859575,
+      "grad_norm": 0.1491909772157669,
+      "learning_rate": 4.932765110981894e-05,
+      "loss": 0.1261,
+      "step": 13600
+    },
+    {
+      "epoch": 1.2204877617352468,
+      "grad_norm": 0.15451796352863312,
+      "learning_rate": 4.929354590919424e-05,
+      "loss": 0.1273,
+      "step": 13800
+    },
+    {
+      "epoch": 1.2381763106109183,
+      "grad_norm": 0.1513613909482956,
+      "learning_rate": 4.9258609357930686e-05,
+      "loss": 0.1264,
+      "step": 14000
+    },
+    {
+      "epoch": 1.2558648594865898,
+      "grad_norm": 0.1397712677717209,
+      "learning_rate": 4.9222842651593736e-05,
+      "loss": 0.1268,
+      "step": 14200
+    },
+    {
+      "epoch": 1.2735534083622615,
+      "grad_norm": 0.13348053395748138,
+      "learning_rate": 4.918624701415763e-05,
+      "loss": 0.1267,
+      "step": 14400
+    },
+    {
+      "epoch": 1.2912419572379332,
+      "grad_norm": 0.15651994943618774,
+      "learning_rate": 4.9148823697963465e-05,
+      "loss": 0.1258,
+      "step": 14600
+    },
+    {
+      "epoch": 1.3089305061136047,
+      "grad_norm": 0.16271665692329407,
+      "learning_rate": 4.9110573983676414e-05,
+      "loss": 0.1258,
+      "step": 14800
+    },
+    {
+      "epoch": 1.3266190549892762,
+      "grad_norm": 0.15141454339027405,
+      "learning_rate": 4.907149918024185e-05,
+      "loss": 0.1252,
+      "step": 15000
+    },
+    {
+      "epoch": 1.344307603864948,
+      "grad_norm": 0.14476826786994934,
+      "learning_rate": 4.903160062484056e-05,
+      "loss": 0.1263,
+      "step": 15200
+    },
+    {
+      "epoch": 1.3619961527406195,
+      "grad_norm": 0.14604584872722626,
+      "learning_rate": 4.8990879682842964e-05,
+      "loss": 0.1267,
+      "step": 15400
+    },
+    {
+      "epoch": 1.3796847016162912,
+      "grad_norm": 0.1551065295934677,
+      "learning_rate": 4.8949337747762465e-05,
+      "loss": 0.1268,
+      "step": 15600
+    },
+    {
+      "epoch": 1.3973732504919627,
+      "grad_norm": 0.15430127084255219,
+      "learning_rate": 4.890697624120767e-05,
+      "loss": 0.1258,
+      "step": 15800
+    },
+    {
+      "epoch": 1.4150617993676344,
+      "grad_norm": 0.1534918248653412,
+      "learning_rate": 4.886379661283379e-05,
+      "loss": 0.1245,
+      "step": 16000
+    },
+    {
+      "epoch": 1.432750348243306,
+      "grad_norm": 0.15945963561534882,
+      "learning_rate": 4.881980034029303e-05,
+      "loss": 0.1251,
+      "step": 16200
+    },
+    {
+      "epoch": 1.4504388971189777,
+      "grad_norm": 0.15190242230892181,
+      "learning_rate": 4.877498892918403e-05,
+      "loss": 0.1246,
+      "step": 16400
+    },
+    {
+      "epoch": 1.4681274459946492,
+      "grad_norm": 0.14410296082496643,
+      "learning_rate": 4.872936391300029e-05,
+      "loss": 0.1251,
+      "step": 16600
+    },
+    {
+      "epoch": 1.485815994870321,
+      "grad_norm": 0.13909801840782166,
+      "learning_rate": 4.868292685307776e-05,
+      "loss": 0.1255,
+      "step": 16800
+    },
+    {
+      "epoch": 1.5035045437459924,
+      "grad_norm": 0.14646555483341217,
+      "learning_rate": 4.8635679338541364e-05,
+      "loss": 0.1243,
+      "step": 17000
+    },
+    {
+      "epoch": 1.521193092621664,
+      "grad_norm": 0.14244422316551208,
+      "learning_rate": 4.858762298625065e-05,
+      "loss": 0.1248,
+      "step": 17200
+    },
+    {
+      "epoch": 1.5388816414973356,
+      "grad_norm": 0.15146219730377197,
+      "learning_rate": 4.853875944074442e-05,
+      "loss": 0.1235,
+      "step": 17400
+    },
+    {
+      "epoch": 1.5565701903730074,
+      "grad_norm": 0.14190447330474854,
+      "learning_rate": 4.848909037418449e-05,
+      "loss": 0.1242,
+      "step": 17600
+    },
+    {
+      "epoch": 1.5742587392486789,
+      "grad_norm": 0.13961580395698547,
+      "learning_rate": 4.8438617486298455e-05,
+      "loss": 0.1235,
+      "step": 17800
+    },
+    {
+      "epoch": 1.5919472881243504,
+      "grad_norm": 0.13622143864631653,
+      "learning_rate": 4.838734250432152e-05,
+      "loss": 0.1231,
+      "step": 18000
+    },
+    {
+      "epoch": 1.609635837000022,
+      "grad_norm": 0.140974760055542,
+      "learning_rate": 4.833526718293736e-05,
+      "loss": 0.1229,
+      "step": 18200
+    },
+    {
+      "epoch": 1.6273243858756938,
+      "grad_norm": 0.1440068930387497,
+      "learning_rate": 4.828239330421815e-05,
+      "loss": 0.1233,
+      "step": 18400
+    },
+    {
+      "epoch": 1.6450129347513653,
+      "grad_norm": 0.13277290761470795,
+      "learning_rate": 4.822872267756351e-05,
+      "loss": 0.122,
+      "step": 18600
+    },
+    {
+      "epoch": 1.6627014836270368,
+      "grad_norm": 0.1427055150270462,
+      "learning_rate": 4.817425713963861e-05,
+      "loss": 0.123,
+      "step": 18800
+    },
+    {
+      "epoch": 1.6803900325027086,
+      "grad_norm": 0.14716538786888123,
+      "learning_rate": 4.8118998554311336e-05,
+      "loss": 0.1235,
+      "step": 19000
+    },
+    {
+      "epoch": 1.6980785813783803,
+      "grad_norm": 0.14058035612106323,
+      "learning_rate": 4.806294881258846e-05,
+      "loss": 0.1219,
+      "step": 19200
+    },
+    {
+      "epoch": 1.7157671302540518,
+      "grad_norm": 0.1426689326763153,
+      "learning_rate": 4.800610983255098e-05,
+      "loss": 0.1223,
+      "step": 19400
+    },
+    {
+      "epoch": 1.7334556791297233,
+      "grad_norm": 0.142822727560997,
+      "learning_rate": 4.7948483559288445e-05,
+      "loss": 0.1217,
+      "step": 19600
+    },
+    {
+      "epoch": 1.751144228005395,
+      "grad_norm": 0.14739733934402466,
+      "learning_rate": 4.7890071964832426e-05,
+      "loss": 0.122,
+      "step": 19800
+    },
+    {
+      "epoch": 1.7688327768810668,
+      "grad_norm": 0.1422308385372162,
+      "learning_rate": 4.7830877048088974e-05,
+      "loss": 0.1209,
+      "step": 20000
+    },
+    {
+      "epoch": 1.7865213257567383,
+      "grad_norm": 0.13312865793704987,
+      "learning_rate": 4.777090083477027e-05,
+      "loss": 0.1209,
+      "step": 20200
+    },
+    {
+      "epoch": 1.8042098746324098,
+      "grad_norm": 0.1410498321056366,
+      "learning_rate": 4.771014537732529e-05,
+      "loss": 0.1202,
+      "step": 20400
+    },
+    {
+      "epoch": 1.8218984235080815,
+      "grad_norm": 0.14671219885349274,
+      "learning_rate": 4.764861275486956e-05,
+      "loss": 0.1195,
+      "step": 20600
+    },
+    {
+      "epoch": 1.839586972383753,
+      "grad_norm": 0.13390739262104034,
+      "learning_rate": 4.758630507311399e-05,
+      "loss": 0.1204,
+      "step": 20800
+    },
+    {
+      "epoch": 1.8572755212594245,
+      "grad_norm": 0.1377139389514923,
+      "learning_rate": 4.7523224464292855e-05,
+      "loss": 0.1194,
+      "step": 21000
+    },
+    {
+      "epoch": 1.8749640701350963,
+      "grad_norm": 0.13953597843647003,
+      "learning_rate": 4.745937308709079e-05,
+      "loss": 0.1196,
+      "step": 21200
+    },
+    {
+      "epoch": 1.892652619010768,
+      "grad_norm": 0.14722049236297607,
+      "learning_rate": 4.739475312656895e-05,
+      "loss": 0.1189,
+      "step": 21400
+    },
+    {
+      "epoch": 1.9103411678864395,
+      "grad_norm": 0.129732146859169,
+      "learning_rate": 4.7329366794090205e-05,
+      "loss": 0.1195,
+      "step": 21600
+    },
+    {
+      "epoch": 1.928029716762111,
+      "grad_norm": 0.13127955794334412,
+      "learning_rate": 4.726321632724346e-05,
+      "loss": 0.1188,
+      "step": 21800
+    },
+    {
+      "epoch": 1.9457182656377827,
+      "grad_norm": 0.1353120505809784,
+      "learning_rate": 4.719630398976714e-05,
+      "loss": 0.1184,
+      "step": 22000
+    },
+    {
+      "epoch": 1.9634068145134544,
+      "grad_norm": 0.14046898484230042,
+      "learning_rate": 4.7128632071471667e-05,
+      "loss": 0.1185,
+      "step": 22200
+    },
+    {
+      "epoch": 1.981095363389126,
+      "grad_norm": 0.13207530975341797,
+      "learning_rate": 4.7060202888161106e-05,
+      "loss": 0.1174,
+      "step": 22400
+    },
+    {
+      "epoch": 1.9987839122647975,
+      "grad_norm": 0.1329745352268219,
+      "learning_rate": 4.6991018781553926e-05,
+      "loss": 0.1181,
+      "step": 22600
+    },
+    {
+      "epoch": 2.0164503504543747,
+      "grad_norm": 0.13534873723983765,
+      "learning_rate": 4.692108211920287e-05,
+      "loss": 0.097,
+      "step": 22800
+    },
+    {
+      "epoch": 2.0341388993300464,
+      "grad_norm": 0.13999226689338684,
+      "learning_rate": 4.685039529441393e-05,
+      "loss": 0.096,
+      "step": 23000
+    },
+    {
+      "epoch": 2.0518274482057177,
+      "grad_norm": 0.14331400394439697,
+      "learning_rate": 4.677896072616444e-05,
+      "loss": 0.0956,
+      "step": 23200
+    },
+    {
+      "epoch": 2.0695159970813894,
+      "grad_norm": 0.1391351968050003,
+      "learning_rate": 4.67067808590203e-05,
+      "loss": 0.0955,
+      "step": 23400
+    },
+    {
+      "epoch": 2.087204545957061,
+      "grad_norm": 0.15080305933952332,
+      "learning_rate": 4.6633858163052324e-05,
+      "loss": 0.0966,
+      "step": 23600
+    },
+    {
+      "epoch": 2.1048930948327325,
+      "grad_norm": 0.14853306114673615,
+      "learning_rate": 4.656019513375171e-05,
+      "loss": 0.0955,
+      "step": 23800
+    },
+    {
+      "epoch": 2.122581643708404,
+      "grad_norm": 0.14308440685272217,
+      "learning_rate": 4.648579429194463e-05,
+      "loss": 0.0959,
+      "step": 24000
+    },
+    {
+      "epoch": 2.140270192584076,
+      "grad_norm": 0.14518137276172638,
+      "learning_rate": 4.641065818370597e-05,
+      "loss": 0.0964,
+      "step": 24200
+    },
+    {
+      "epoch": 2.1579587414597476,
+      "grad_norm": 0.14385883510112762,
+      "learning_rate": 4.6334789380272235e-05,
+      "loss": 0.0966,
+      "step": 24400
+    },
+    {
+      "epoch": 2.175647290335419,
+      "grad_norm": 0.14189130067825317,
+      "learning_rate": 4.625819047795349e-05,
+      "loss": 0.0969,
+      "step": 24600
+    },
+    {
+      "epoch": 2.1933358392110907,
+      "grad_norm": 0.14711739122867584,
+      "learning_rate": 4.6180864098044584e-05,
+      "loss": 0.0967,
+      "step": 24800
+    },
+    {
+      "epoch": 2.2110243880867624,
+      "grad_norm": 0.14622507989406586,
+      "learning_rate": 4.610281288673539e-05,
+      "loss": 0.0967,
+      "step": 25000
+    },
+    {
+      "epoch": 2.228712936962434,
+      "grad_norm": 0.1425817757844925,
+      "learning_rate": 4.6024039515020276e-05,
+      "loss": 0.0981,
+      "step": 25200
+    },
+    {
+      "epoch": 2.2464014858381054,
+      "grad_norm": 0.1472136527299881,
+      "learning_rate": 4.5944546678606706e-05,
+      "loss": 0.0993,
+      "step": 25400
+    },
+    {
+      "epoch": 2.264090034713777,
+      "grad_norm": 0.14232853055000305,
+      "learning_rate": 4.586433709782296e-05,
+      "loss": 0.0985,
+      "step": 25600
+    },
+    {
+      "epoch": 2.281778583589449,
+      "grad_norm": 0.1511968970298767,
+      "learning_rate": 4.578341351752511e-05,
+      "loss": 0.098,
+      "step": 25800
+    },
+    {
+      "epoch": 2.2994671324651206,
+      "grad_norm": 0.14743036031723022,
+      "learning_rate": 4.570177870700298e-05,
+      "loss": 0.0974,
+      "step": 26000
+    },
+    {
+      "epoch": 2.317155681340792,
+      "grad_norm": 0.1461828500032425,
+      "learning_rate": 4.561943545988548e-05,
+      "loss": 0.0969,
+      "step": 26200
+    },
+    {
+      "epoch": 2.3348442302164636,
+      "grad_norm": 0.1614687293767929,
+      "learning_rate": 4.5536386594044956e-05,
+      "loss": 0.0988,
+      "step": 26400
+    },
+    {
+      "epoch": 2.3525327790921353,
+      "grad_norm": 0.1400858461856842,
+      "learning_rate": 4.5452634951500745e-05,
+      "loss": 0.0969,
+      "step": 26600
+    },
+    {
+      "epoch": 2.3702213279678066,
+      "grad_norm": 0.14338116347789764,
+      "learning_rate": 4.536818339832197e-05,
+      "loss": 0.0961,
+      "step": 26800
+    },
+    {
+      "epoch": 2.3879098768434783,
+      "grad_norm": 0.1501627415418625,
+      "learning_rate": 4.528303482452943e-05,
+      "loss": 0.0969,
+      "step": 27000
+    },
+    {
+      "epoch": 2.40559842571915,
+      "grad_norm": 0.14346472918987274,
+      "learning_rate": 4.519719214399667e-05,
+      "loss": 0.0958,
+      "step": 27200
+    },
+    {
+      "epoch": 2.423286974594822,
+      "grad_norm": 0.1539745032787323,
+      "learning_rate": 4.5110658294350326e-05,
+      "loss": 0.0966,
+      "step": 27400
+    },
+    {
+      "epoch": 2.4409755234704935,
+      "grad_norm": 0.1377008557319641,
+      "learning_rate": 4.502343623686956e-05,
+      "loss": 0.0971,
+      "step": 27600
+    },
+    {
+      "epoch": 2.458664072346165,
+      "grad_norm": 0.1407323032617569,
+      "learning_rate": 4.493552895638472e-05,
+      "loss": 0.0974,
+      "step": 27800
+    },
+    {
+      "epoch": 2.4763526212218365,
+      "grad_norm": 0.1503271609544754,
+      "learning_rate": 4.48469394611752e-05,
+      "loss": 0.0975,
+      "step": 28000
+    },
+    {
+      "epoch": 2.4940411700975083,
+      "grad_norm": 0.14528031647205353,
+      "learning_rate": 4.475767078286652e-05,
+      "loss": 0.0974,
+      "step": 28200
+    },
+    {
+      "epoch": 2.5117297189731795,
+      "grad_norm": 0.14073877036571503,
+      "learning_rate": 4.466772597632654e-05,
+      "loss": 0.0963,
+      "step": 28400
+    },
+    {
+      "epoch": 2.5294182678488513,
+      "grad_norm": 0.14889408648014069,
+      "learning_rate": 4.457710811956094e-05,
+      "loss": 0.0963,
+      "step": 28600
+    },
+    {
+      "epoch": 2.547106816724523,
+      "grad_norm": 0.14946310222148895,
+      "learning_rate": 4.4485820313607906e-05,
+      "loss": 0.0967,
+      "step": 28800
+    },
+    {
+      "epoch": 2.5647953656001947,
+      "grad_norm": 0.1508057564496994,
+      "learning_rate": 4.4393865682431955e-05,
+      "loss": 0.0973,
+      "step": 29000
+    },
+    {
+      "epoch": 2.5824839144758664,
+      "grad_norm": 0.1469687819480896,
+      "learning_rate": 4.4301247372817077e-05,
+      "loss": 0.0971,
+      "step": 29200
+    },
+    {
+      "epoch": 2.6001724633515377,
+      "grad_norm": 0.14045564830303192,
+      "learning_rate": 4.420796855425905e-05,
+      "loss": 0.0959,
+      "step": 29400
+    },
+    {
+      "epoch": 2.6178610122272095,
+      "grad_norm": 0.14008161425590515,
+      "learning_rate": 4.411403241885693e-05,
+      "loss": 0.0966,
+      "step": 29600
+    },
+    {
+      "epoch": 2.635549561102881,
+      "grad_norm": 0.15664730966091156,
+      "learning_rate": 4.4019442181203884e-05,
+      "loss": 0.0969,
+      "step": 29800
+    },
+    {
+      "epoch": 2.6532381099785525,
+      "grad_norm": 0.146159827709198,
+      "learning_rate": 4.3924201078277105e-05,
+      "loss": 0.0969,
+      "step": 30000
+    },
+    {
+      "epoch": 2.670926658854224,
+      "grad_norm": 0.14852313697338104,
+      "learning_rate": 4.382831236932711e-05,
+      "loss": 0.0955,
+      "step": 30200
+    },
+    {
+      "epoch": 2.688615207729896,
+      "grad_norm": 0.15001444518566132,
+      "learning_rate": 4.3731779335766154e-05,
+      "loss": 0.0959,
+      "step": 30400
+    },
+    {
+      "epoch": 2.7063037566055677,
+      "grad_norm": 0.14691965281963348,
+      "learning_rate": 4.363460528105597e-05,
+      "loss": 0.0957,
+      "step": 30600
+    },
+    {
+      "epoch": 2.723992305481239,
+      "grad_norm": 0.14318469166755676,
+      "learning_rate": 4.35367935305947e-05,
+      "loss": 0.0967,
+      "step": 30800
+    },
+    {
+      "epoch": 2.7416808543569107,
+      "grad_norm": 0.14508825540542603,
+      "learning_rate": 4.34383474316031e-05,
+      "loss": 0.0951,
+      "step": 31000
+    },
+    {
+      "epoch": 2.7593694032325824,
+      "grad_norm": 0.15151719748973846,
+      "learning_rate": 4.333927035301001e-05,
+      "loss": 0.0958,
+      "step": 31200
+    },
+    {
+      "epoch": 2.7770579521082537,
+      "grad_norm": 0.1437998265028,
+      "learning_rate": 4.3239565685337044e-05,
+      "loss": 0.0955,
+      "step": 31400
+    },
+    {
+      "epoch": 2.7947465009839254,
+      "grad_norm": 0.14122720062732697,
+      "learning_rate": 4.3139236840582575e-05,
+      "loss": 0.0951,
+      "step": 31600
+    },
+    {
+      "epoch": 2.812435049859597,
+      "grad_norm": 0.1338931769132614,
+      "learning_rate": 4.303828725210498e-05,
+      "loss": 0.0959,
+      "step": 31800
+    },
+    {
+      "epoch": 2.830123598735269,
+      "grad_norm": 0.144461989402771,
+      "learning_rate": 4.293672037450512e-05,
+      "loss": 0.0951,
+      "step": 32000
+    },
+    {
+      "epoch": 2.8478121476109406,
+      "grad_norm": 0.13668565452098846,
+      "learning_rate": 4.2834539683508166e-05,
+      "loss": 0.0959,
+      "step": 32200
+    },
+    {
+      "epoch": 2.865500696486612,
+      "grad_norm": 0.15980184078216553,
+      "learning_rate": 4.27317486758446e-05,
+      "loss": 0.0951,
+      "step": 32400
+    },
+    {
+      "epoch": 2.8831892453622836,
+      "grad_norm": 0.14222297072410583,
+      "learning_rate": 4.262835086913058e-05,
+      "loss": 0.0948,
+      "step": 32600
+    },
+    {
+      "epoch": 2.9008777942379553,
+      "grad_norm": 0.1509125828742981,
+      "learning_rate": 4.25243498017476e-05,
+      "loss": 0.0949,
+      "step": 32800
+    },
+    {
+      "epoch": 2.9185663431136266,
+      "grad_norm": 0.14582620561122894,
+      "learning_rate": 4.241974903272132e-05,
+      "loss": 0.0943,
+      "step": 33000
+    },
+    {
+      "epoch": 2.9362548919892983,
+      "grad_norm": 0.130904421210289,
+      "learning_rate": 4.231455214159985e-05,
+      "loss": 0.0945,
+      "step": 33200
+    },
+    {
+      "epoch": 2.95394344086497,
+      "grad_norm": 0.1373138129711151,
+      "learning_rate": 4.220876272833121e-05,
+      "loss": 0.0936,
+      "step": 33400
+    },
+    {
+      "epoch": 2.971631989740642,
+      "grad_norm": 0.14548689126968384,
+      "learning_rate": 4.210238441314017e-05,
+      "loss": 0.0934,
+      "step": 33600
+    },
+    {
+      "epoch": 2.9893205386163135,
+      "grad_norm": 0.13954471051692963,
+      "learning_rate": 4.199542083640432e-05,
+      "loss": 0.0937,
+      "step": 33800
+    },
+    {
+      "epoch": 3.0069869768058903,
+      "grad_norm": 0.14478568732738495,
+      "learning_rate": 4.188787565852952e-05,
+      "loss": 0.0847,
+      "step": 34000
+    },
+    {
+      "epoch": 3.024675525681562,
+      "grad_norm": 0.15503637492656708,
+      "learning_rate": 4.177975255982463e-05,
+      "loss": 0.0718,
+      "step": 34200
+    },
+    {
+      "epoch": 3.0423640745572333,
+      "grad_norm": 0.16748104989528656,
+      "learning_rate": 4.1671055240375575e-05,
+      "loss": 0.0714,
+      "step": 34400
+    },
+    {
+      "epoch": 3.060052623432905,
+      "grad_norm": 0.15309254825115204,
+      "learning_rate": 4.156178741991872e-05,
+      "loss": 0.0724,
+      "step": 34600
+    },
+    {
+      "epoch": 3.077741172308577,
+      "grad_norm": 0.15351246297359467,
+      "learning_rate": 4.1451952837713556e-05,
+      "loss": 0.0722,
+      "step": 34800
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 113060,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0180563807244386e+20,
+  "train_batch_size": 9,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e0ca058fbdf0074da43c08f58e1ce54d3cdb66ab040c892c722ccc273fd1644
+size 7761

zero_to_fp32.py

@@ -0,0 +1,760 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example:
+#   python zero_to_fp32.py . output_dir/
+#   or
+#   python zero_to_fp32.py . output_dir/ --safe_serialization
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+import gc
+import json
+import numpy as np
+from tqdm import tqdm
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device, weights_only=False)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+    total_files = len(files)
+    state_dicts = []
+    for f in tqdm(files, desc='Loading checkpoint shards'):
+        state_dict = torch.load(f, map_location=device, mmap=True, weights_only=False)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+class GatheredTensor:
+    """
+    A pseudo tensor that collects partitioned weights.
+    It is more memory efficient when there are multiple groups.
+    """
+
+    def __init__(self, flat_groups, flat_groups_offset, offset, partitioned_numel, shape):
+        self.flat_groups = flat_groups
+        self.flat_groups_offset = flat_groups_offset
+        self.offset = offset
+        self.partitioned_numel = partitioned_numel
+        self.shape = shape
+        self.dtype = self.flat_groups[0][0].dtype
+
+    def contiguous(self):
+        """
+        Merge partitioned weights from flat_groups into a single tensor.
+        """
+        end_idx = self.offset + self.partitioned_numel
+        world_size = len(self.flat_groups)
+        pad_flat_param_chunks = []
+
+        for rank_i in range(world_size):
+            # for each rank, we need to collect weights from related group/groups
+            flat_groups_at_rank_i = self.flat_groups[rank_i]
+            start_group_id = None
+            end_group_id = None
+            for group_id in range(len(self.flat_groups_offset)):
+                if self.flat_groups_offset[group_id] <= self.offset < self.flat_groups_offset[group_id + 1]:
+                    start_group_id = group_id
+                if self.flat_groups_offset[group_id] < end_idx <= self.flat_groups_offset[group_id + 1]:
+                    end_group_id = group_id
+                    break
+            # collect weights from related group/groups
+            for group_id in range(start_group_id, end_group_id + 1):
+                flat_tensor = flat_groups_at_rank_i[group_id]
+                start_offset = self.offset - self.flat_groups_offset[group_id]
+                end_offset = min(end_idx, self.flat_groups_offset[group_id + 1]) - self.flat_groups_offset[group_id]
+                pad_flat_param_chunks.append(flat_tensor[start_offset:end_offset])
+
+        # collect weights from all ranks
+        pad_flat_param = torch.cat(pad_flat_param_chunks, dim=0)
+        param = pad_flat_param[:self.shape.numel()].view(self.shape).contiguous()
+        return param
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = sum([flat_group.numel() for flat_group in fp32_flat_groups[0]]) * world_size
+
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    flat_groups_offset = [0] + list(np.cumsum([flat_tensor.numel() for flat_tensor in fp32_flat_groups[0]]))
+    for name, shape in tqdm(param_shapes.items(), desc='Gathering sharded weights'):
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # memory efficient tensor
+        tensor = GatheredTensor(fp32_flat_groups, flat_groups_offset, offset, partitioned_numel, shape)
+        state_dict[name] = tensor
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def to_torch_tensor(state_dict, return_empty_tensor=False):
+    """
+    Convert state_dict of GatheredTensor to torch tensor
+    """
+    torch_state_dict = {}
+    converted_tensors = {}
+    for name, tensor in state_dict.items():
+        tensor_id = id(tensor)
+        if tensor_id in converted_tensors:  # shared tensors
+            shared_tensor = torch_state_dict[converted_tensors[tensor_id]]
+            torch_state_dict[name] = shared_tensor
+        else:
+            converted_tensors[tensor_id] = name
+            if return_empty_tensor:
+                torch_state_dict[name] = torch.empty(tensor.shape, dtype=tensor.dtype)
+            else:
+                torch_state_dict[name] = tensor.contiguous()
+    return torch_state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                             tag=None,
+                                             exclude_frozen_parameters=False,
+                                             lazy_mode=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+        - ``lazy_mode``: get state_dict in lazy mode. It returns a dict of pesduo tensor instead of torch tensor, which is more memory efficient.
+          Convert the pesduo tensor to torch tensor by ``.contiguous()``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    Note: the above usage may not work if your application doesn't have sufficient free CPU memory.
+    You may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint. Or you can load state_dict in lazy mode ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, lazy_mode=True) # not on cpu
+        for name, lazy_tensor in state_dict.item():
+            tensor = lazy_tensor.contiguous()  # to cpu
+            print(name, tensor)
+            # del tensor to release memory if it no longer in use
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    state_dict = _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+    if lazy_mode:
+        return state_dict
+    else:
+        return to_torch_tensor(state_dict)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir,
+                                               output_dir,
+                                               max_shard_size="5GB",
+                                               safe_serialization=False,
+                                               tag=None,
+                                               exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_dir``: directory to the pytorch fp32 state_dict output files
+        - ``max_shard_size``: the maximum size for a checkpoint before being sharded, default value is 5GB
+        - ``safe_serialization``:  whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    # Dependency pre-check
+    if safe_serialization:
+        try:
+            from safetensors.torch import save_file
+        except ImportError:
+            print('If you want to use `safe_serialization`, please `pip install safetensors`')
+            raise
+    if max_shard_size is not None:
+        try:
+            from huggingface_hub import split_torch_state_dict_into_shards
+        except ImportError:
+            print('If you want to use `max_shard_size`, please `pip install huggingface_hub`')
+            raise
+
+    # Convert zero checkpoint to state_dict
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir,
+                                                          tag,
+                                                          exclude_frozen_parameters,
+                                                          lazy_mode=True)
+
+    # Shard the model if it is too big.
+    weights_name = "model.safetensors" if safe_serialization else "pytorch_model.bin"
+    if max_shard_size is not None:
+        filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(".safetensors", "{suffix}.safetensors")
+        # an memory-efficient approach for sharding
+        empty_state_dict = to_torch_tensor(state_dict, return_empty_tensor=True)
+        state_dict_split = split_torch_state_dict_into_shards(empty_state_dict,
+                                                              filename_pattern=filename_pattern,
+                                                              max_shard_size=max_shard_size)
+    else:
+        from collections import namedtuple
+        StateDictSplit = namedtuple("StateDictSplit", ["is_sharded", "filename_to_tensors"])
+        state_dict_split = StateDictSplit(is_sharded=False,
+                                          filename_to_tensors={weights_name: list(state_dict.keys())})
+
+    # Save the model by shard
+    os.makedirs(output_dir, exist_ok=True)
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+    for shard_file, tensors in tqdm(filename_to_tensors, desc="Saving checkpoint shards"):
+        shard_state_dict = {tensor_name: state_dict[tensor_name] for tensor_name in tensors}
+        shard_state_dict = to_torch_tensor(shard_state_dict)
+        output_path = os.path.join(output_dir, shard_file)
+        if safe_serialization:
+            save_file(shard_state_dict, output_path, metadata={"format": "pt"})
+        else:
+            torch.save(shard_state_dict, output_path)
+        # release the memory of current shard
+        for tensor_name in list(shard_state_dict.keys()):
+            del state_dict[tensor_name]
+            del shard_state_dict[tensor_name]
+        del shard_state_dict
+        gc.collect()
+
+    # Save index if sharded
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+        save_index_file = "model.safetensors.index.json" if safe_serialization else "pytorch_model.bin.index.json"
+        save_index_file = os.path.join(output_dir, save_index_file)
+        with open(save_index_file, "w", encoding="utf-8") as f:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            f.write(content)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument("output_dir",
+                        type=str,
+                        help="directory to the pytorch fp32 state_dict output files"
+                        "(e.g. path/checkpoint-12-output/)")
+    parser.add_argument(
+        "--max_shard_size",
+        type=str,
+        default="5GB",
+        help="The maximum size for a checkpoint before being sharded. Checkpoints shard will then be each of size"
+        "lower than this size. If expressed as a string, needs to be digits followed by a unit (like `5MB`"
+        "We default it to 5GB in order for models to be able to run easily on free-tier google colab instances"
+        "without CPU OOM issues.")
+    parser.add_argument(
+        "--safe_serialization",
+        default=False,
+        action='store_true',
+        help="Whether to save the model using `safetensors` or the traditional PyTorch way (that uses `pickle`).")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_dir,
+                                               max_shard_size=args.max_shard_size,
+                                               safe_serialization=args.safe_serialization,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)