Functions — github.com/amazon-science/controlling-llm-memorization

CodeHub MCPcopy Create free account

hub / github.com/amazon-science/controlling-llm-memorization / functions

Functions12 in github.com/amazon-science/controlling-llm-memorization

⨍ Functions 12 ◇ Types & classes 1

↓ 1 callersMethodinitialize_embedding

initializes learned embedding Args: same as __init__ Returns: torch.float: initialized using original scheme

src/my_utils.py:129

↓ 1 callersFunctionmain

src/baseline.py:22

↓ 1 callersFunctionmain

src/promptLearn_defense.py:23

↓ 1 callersFunctionmain

src/promptLearn_attack.py:23

appends learned embedding to Args: wte (nn.Embedding): original transformer word embedding n_tokens (int, optional):

src/my_utils.py:108

Functioncompute_reconstruct_rate

compute fractional and exact reconstruction rates

src/my_utils.py:16

Functionevaluate

get inference loss on supplied data loader

src/my_utils.py:43

Functionevaluate_distributed

get inference loss on supplied data loader (for distributed training)

src/my_utils.py:82

run forward pass Args: tokens (torch.long): input tokens before encoding Returns: torch.float: encoding of t

src/my_utils.py:148

Functiongenerate_suffixes

generate suffixes from the supplied data loader

src/my_utils.py:24

Functiongenerate_suffixes_distributed

generate suffixes from the supplied data loader (for distributed training)

src/my_utils.py:60

Functionload_prompts

Loads prompts from the file pointed to prompt_path

src/my_utils.py:11