INDEX

Explanations

statistical or survey-related terms such as "respondents" and percentages

oai_token-act-pair · gpt-3.5-turbo

instances of the word "respondents" and related references to survey participants

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

Cop

-0.75

 Sabb

-0.74

cop

-0.72

Cop

-0.68

 Cart

-0.67

 Sabha

-0.65

 Fury

-0.65

 slic

-0.65

 marrow

-0.64

 Revival

-0.64

POSITIVE LOGITS

 respondents

0.94

uggest

0.89

 answ

0.86

 respondent

0.85

 interviewed

0.80

 surveyed

0.79

 answered

0.77

omnia

0.74

 fielded

0.73

ixty

0.72

Activations Density 0.031%

statistical or survey-related terms such as "respondents" and percentages

instances of the word "respondents" and related references to survey participants

No Comments

No Known Activations