INDEX

Explanations

terms related to guidance or instructions given by authorities

oai_token-act-pair · gpt-3.5-turbo

phrases that mention guidance or directives from authorities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ols

-0.71

itol

-0.68

ells

-0.66

tones

-0.66

 forfeit

-0.65

fer

-0.64

asts

-0.64

oreal

-0.63

itals

-0.63

 Alive

-0.63

POSITIVE LOGITS

 guidance

1.06

 counselors

1.05

 counselor

0.93

 Advice

0.90

 Guid

0.82

arity

0.82

 advising

0.81

tips

0.77

 guidelines

0.77

 guides

0.76

Activations Density 0.022%

terms related to guidance or instructions given by authorities

phrases that mention guidance or directives from authorities

No Comments

No Known Activations