Functions — github.com/AnswerDotAI/ModernBERT

hub / github.com/AnswerDotAI/ModernBERT / functions

Functions560 in github.com/AnswerDotAI/ModernBERT

⨍ Functions 560◇ Types & classes 136↳ Endpoints 2

Returns a batch of packed sequences with its cumulative seq length information. Or else, returns None if it cannot build a full outg

src/sequence_packer.py:348

↓ 1 callersFunction_est_progress_denominator

( total_samples: int, chars_per_sample: int, chars_per_token: int, mode: ConcatMode, max_length: int )

src/convert_dataset.py:305

↓ 1 callersMethod_fill_buffer

Refills the internal buffer. - max_items_to_add: an amount less than or equal to the number of items to add Returns: the nu

src/sequence_packer.py:196

↓ 1 callersMethod_finish_setup

(self, t_max: Time[int], t_warmup: Time[int], t_cosine: Time[int], t_cooldown: Time[int])

src/scheduler.py:240

↓ 1 callersFunction_foreach_stableadamw

( params: list[Tensor], grads: list[Tensor], exp_avgs: list[Tensor], exp_avg_sqs: list[Tensor]

src/optimizer.py:250

↓ 1 callersMethod_generate_batches

Generates batches of packed sequences. The returned generator's iterator will always, when next() is called on it, either:

src/sequence_packer.py:235

↓ 1 callersMethod_init_group

( self, group: dict[str, Any], params: list[Tensor], grads: list[Tensor],

src/optimizer.py:104

↓ 1 callersMethod_init_state

(self, group: dict[str, Any], state: dict[Tensor, Any], param: Tensor)

src/optimizer.py:89

↓ 1 callersMethod_init_weights

(self, reset_params: bool = False)

src/bert_layers/layers.py:601

↓ 1 callersMethod_init_weights

(self, reset_params: bool = False)

src/bert_layers/mlp.py:81

↓ 1 callersMethod_init_weights

(self, reset_params: bool = False)

src/bert_layers/embeddings.py:113

↓ 1 callersMethod_init_weights

(self, reset_params: bool = False)

src/bert_layers/model.py:736

↓ 1 callersMethod_init_weights

(self, reset_params: bool = False)

src/bert_layers/model.py:771

↓ 1 callersMethod_init_weights

(self, module: Optional[nn.Module] = None, reset_params: Optional[bool] = None)

src/bert_layers/model.py:1025

↓ 1 callersMethod_init_weights

(self, module: Optional[nn.Module] = None, reset_params: Optional[bool] = None)

src/bert_layers/model.py:1268

↓ 1 callersMethod_init_weights

(self, module: Optional[nn.Module] = None, reset_params: Optional[bool] = None)

src/bert_layers/model.py:1403

↓ 1 callersFunction_make_passage

(doc)

_colbert.py:14

↓ 1 callersMethod_norm

Apply the RMSNorm normalization to the input tensor. Args: x (torch.Tensor): The input tensor. Returns:

src/bert_layers/normalization.py:44

↓ 1 callersMethod_packing_efficiency

(self, state: State)

src/callbacks/packing_efficiency.py:25

↓ 1 callersFunction_print_table

Pretty prints a table given a results dictionary.

eval.py:456

↓ 1 callersFunction_print_table

Pretty prints a table given a results dictionary.

glue.py:390

↓ 1 callersMethod_read_binary_tokenized_sample

(self, sample: BatchEncoding)

src/text_data.py:212

↓ 1 callersMethod_reset_state

(self)

src/sequence_packer.py:174

↓ 1 callersFunction_setup_gpu_queue

Returns a queue with [0, 1, .. num_gpus].

eval.py:270

↓ 1 callersFunction_setup_gpu_queue

Returns a queue with [0, 1, .. num_gpus].

glue.py:220

↓ 1 callersFunction_tile_1d

(pretrained_weights: torch.Tensor, new_weights: torch.Tensor, mode: TileMode)

src/bert_layers/initialization.py:200

↓ 1 callersFunction_tile_2d

(pretrained_weights: torch.Tensor, new_weights: torch.Tensor, mode: TileMode)

src/bert_layers/initialization.py:232

↓ 1 callersMethod_tokenize

(self, text_sample)

src/text_data.py:205

↓ 1 callersMethod_tokenize

(self, text_sample)

src/text_data.py:483

↓ 1 callersMethodapply

(self, event: Event, state: State, logger: Logger)

src/algorithms/rope_schedule.py:68

↓ 1 callersFunctionbenchmark_inference

(model, dataloader, num_warmup_batches=10, gpu_idx=0)

benchmark.py:190

↓ 1 callersFunctionbenchmark_training

(model, dataloader, num_warmup_batches=10, gpu_idx=0)

benchmark.py:134

↓ 1 callersFunctionbuild_algorithm

(name, kwargs)

eval.py:81

↓ 1 callersFunctionbuild_algorithm

(name, kwargs)

glue.py:73

↓ 1 callersFunctionbuild_algorithm

(name, kwargs)

sequence_classification.py:90

↓ 1 callersFunctionbuild_algorithm

(name, kwargs)

main.py:124

↓ 1 callersFunctionbuild_callback

(name, kwargs)

eval.py:92

↓ 1 callersFunctionbuild_callback

(name, kwargs)

glue.py:84

↓ 1 callersFunctionbuild_callback

(name, kwargs)

sequence_classification.py:101

↓ 1 callersFunctionbuild_callback

(name, kwargs)

main.py:151

↓ 1 callersFunctionbuild_colbert_index

( dataset_name: str, model_name_or_path: str, checkpoint_path: str, collection: list[str],

src/colbert_beir/index_and_score.py:11

↓ 1 callersFunctionbuild_dataloader

(dataset, batch_size)

src/convert_dataset.py:315

↓ 1 callersFunctionbuild_hf_dataset

Build an IterableDataset over the HF C4 or pile source data. Args: dataset_name (str): Dataset name split (str): Split name.

src/convert_dataset.py:247

↓ 1 callersFunctionbuild_logger

(name, kwargs)

eval.py:112

↓ 1 callersFunctionbuild_logger

(name, kwargs)

glue.py:104

↓ 1 callersFunctionbuild_logger

(name, kwargs)

sequence_classification.py:120

↓ 1 callersFunctionbuild_logger

(name, kwargs)

main.py:181

↓ 1 callersFunctionbuild_model

(cfg: DictConfig, num_labels: int, multiple_choice: bool = False, **kwargs)

eval.py:184

↓ 1 callersFunctionbuild_model

( cfg: DictConfig, num_labels: int, multiple_choice: bool = False, **kwargs )

glue.py:126

↓ 1 callersFunctionbuild_model

(cfg: DictConfig)

sequence_classification.py:210

↓ 1 callersFunctionbuild_no_streaming_dataset

( cfg: DictConfig, tokenizer: Tokenizer, pad_sequences: bool = True, )

src/text_data.py:344

↓ 1 callersFunctionbuild_optimizer

(cfg, model)

eval.py:132

↓ 1 callersFunctionbuild_optimizer

(cfg, model)

sequence_classification.py:140

↓ 1 callersFunctionbuild_optimizer

(cfg, model)

main.py:215

↓ 1 callersFunctionbuild_scheduler

(cfg)

eval.py:119

↓ 1 callersFunctionbuild_scheduler

(cfg)

glue.py:111

↓ 1 callersFunctionbuild_scheduler

(cfg)

sequence_classification.py:127

↓ 1 callersFunctionbuild_scheduler

(cfg)

main.py:188

↓ 1 callersFunctionbuild_streaming_dataset

( cfg: DictConfig, tokenizer: Tokenizer, device_batch_size: int, )

src/text_data.py:293

↓ 1 callersFunctionbuild_text_dataloader

( cfg: DictConfig, tokenizer: Tokenizer, device_batch_size: int, device_microbatch_size: int,

src/text_data.py:358

↓ 1 callersFunctionbuild_tokenizer

( om_tokenizer_config: DictConfig, )

src/text_data.py:75

↓ 1 callersFunctioncolbert_score

( model_name_or_path: str, dataset_name: str, dataset: ir_datasets.Dataset, int2docid: dict[in

src/colbert_beir/index_and_score.py:34

↓ 1 callersFunctioncolbert_train

(model_name_or_path: str, train_params: dict, n_gpu: int, data_path: str)

src/colbert_beir/train.py:6

↓ 1 callersMethodcompiled_head

(self, output: torch.Tensor)

src/bert_layers/model.py:1093

↓ 1 callersMethodcompiled_mlp

(self, hidden_states: torch.Tensor)

src/bert_layers/layers.py:303

↓ 1 callersMethodcompute

(self)

src/evals/superglue_jobs.py:319

↓ 1 callersMethodcompute_final

(self)

src/evals/superglue_jobs.py:322

↓ 1 callersFunctioncreate_all_datasets

(tokenizer, num_samples=8192)

efficiency/multiprocess_bench.py:78

↓ 1 callersFunctioncreate_dummy_data

(num_samples, seq_length, vocab_size, model_type)

benchmark.py:239

↓ 1 callersFunctioncreate_fixed_long_dataset

(tokenizer, num_samples=8192)

efficiency/multiprocess_bench.py:24

↓ 1 callersFunctioncreate_fixed_short_dataset

(tokenizer, num_samples=8192)

efficiency/multiprocess_bench.py:16

↓ 1 callersFunctioncreate_symlink_for_newest_checkpoint

Create a symlink to the newest checkpoint file if 'latest-rank0.pt' does not exist.

run_evals.py:332

↓ 1 callersFunctioncreate_synthetic_text_dataset

(n_samples: int = 16)

tests/test_utils.py:25

↓ 1 callersFunctioncreate_variable_long_dataset

(tokenizer, num_samples=8192)

efficiency/multiprocess_bench.py:55

↓ 1 callersFunctioncreate_variable_short_dataset

(tokenizer, num_samples=8192)

efficiency/multiprocess_bench.py:32

↓ 1 callersFunctiondownload_hub_files

Download specific files or the entire repository from a Hugging Face Hub repository.

run_evals.py:541

↓ 1 callersFunctiondownload_starting_checkpoint

Downloads the pretrained checkpoints to start from. Currently only supports S3 and URLs

eval.py:243

↓ 1 callersFunctiondownload_starting_checkpoint

Downloads the pretrained checkpoints to start from. Currently only supports S3 and URLs

glue.py:187

↓ 1 callersFunctionfind_best_fit

(remaining_spaces, seq_len)

src/sequence_packer.py:369

↓ 1 callersFunctionfind_checkpoint_file

(file_path: str, repo_files: List[str])

run_evals.py:518

↓ 1 callersFunctionflatten_config

(cfg)

generate_eval_config.py:141

↓ 1 callersFunctionformat_job_name

Formats the job name for pretty printing.

eval.py:450

↓ 1 callersFunctionformat_job_name

Formats the job name for pretty printing.

glue.py:384

↓ 1 callersMethodforward

Get just the values of `input` which are at `indices`. Arguments: ctx: the autograd context object input: (b, ...) 2+

src/bert_padding.py:22

↓ 1 callersFunctiongenerate_cos_sin

(seqlen, rotary_dim, device, dtype)

tests/test_rotary.py:27

↓ 1 callersFunctiongenerate_eval_configs

Generate evaluation configs for each checkpoint.

run_evals.py:383

↓ 1 callersFunctiongenerate_samples

Generator over samples of a dataloader. Args: loader (DataLoader): A dataloader emitting batches like {key: [sample0_bytes, sample1_bytes,

src/convert_dataset.py:337

↓ 1 callersFunctiongenerate_seqlen_offsets

(seqlen_offsets_type, batch_size, seqlen, device)

tests/test_rotary.py:35

↓ 1 callersFunctionget_base_folder

(artifact_name)

download_artifacts_from_wandb.py:16

↓ 1 callersFunctionget_checkpoint_name_from_path

To go from checkpoint name to path, replace | with /

eval.py:238

↓ 1 callersFunctionget_checkpoint_name_from_path

To go from checkpoint name to path, replace | with /

glue.py:182

↓ 1 callersFunctionget_counts

(args)

src/data/get_counts_from_hf.py:59

↓ 1 callersFunctionget_counts_for_repo

(repo, args)

src/data/get_counts_from_hf.py:25

↓ 1 callersFunctionget_embedding_layer

(config: FlexBertConfig)

src/bert_layers/embeddings.py:210

↓ 1 callersFunctionget_encoder_layer

(config: FlexBertConfig)

src/bert_layers/layers.py:693

↓ 1 callersFunctionget_gpu_memory_usage

Get memory usage for a specific GPU.

run_evals.py:97

↓ 1 callersFunctionget_loss_fn

(config: FlexBertConfig)

src/bert_layers/loss.py:23

↓ 1 callersFunctionget_model_artifacts

(api, entity, project)

download_artifacts_from_wandb.py:11

↓ 1 callersFunctionget_model_defaults

(model_size: ModelSize)

generate_eval_config.py:46

↓ 1 callersMethodget_sequence_id_from_batch

(self, batch: Dict[str, torch.Tensor])

src/text_data.py:280

← previous next →101–200 of 560, ranked by callers