INDEX

Explanations

instances where someone or something is "helped" by someone else

oai_token-act-pair · gpt-3.5-turbo

instances of the word "helped" in various contexts

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

é¾

-0.72

meat

-0.65

 shenan

-0.63

bi

-0.63

Pict

-0.62

posure

-0.61

 Viet

-0.61

Compar

-0.60

NF

-0.60

EVA

-0.58

POSITIVE LOGITS

 usher

0.85

fully

0.81

 oversee

0.78

 facilitate

0.78

uggle

0.76

 organise

0.74

 propel

0.72

oise

0.71

meet

0.70

lead

0.70

Activations Density 0.052%

instances where someone or something is "helped" by someone else

instances of the word "helped" in various contexts

No Comments

No Known Activations