INDEX

Explanations

references to reputation or reputational damage

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reputation" and related phrases indicating perceived credibility or standing

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

cise

-0.89

vention

-0.78

tein

-0.75

pheus

-0.74

nes

-0.70

isting

-0.69

early

-0.68

err

-0.68

scl

-0.67

eping

-0.67

POSITIVE LOGITS

 reputation

1.01

 tremend

0.91

 veter

0.88

 tarn

0.86

©¶æ¥µ

0.82

 internationally

0.80

MPG

0.74

 bearer

0.72

 liar

0.71

 behavi

0.71

Activations Density 0.032%

references to reputation or reputational damage

instances of the word "reputation" and related phrases indicating perceived credibility or standing

No Comments

No Known Activations