INDEX

Explanations

references to superlatives or extremes

oai_token-act-pair · gpt-3.5-turbo

instances of the word "biggest" indicating significant or noteworthy events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ences

-0.77

heid

-0.75

ttes

-0.75

onto

-0.72

ious

-0.71

jad

-0.71

anism

-0.71

actions

-0.70

ENC

-0.70

isson

-0.69

POSITIVE LOGITS

gest

0.98

 culprit

0.93

 obstacle

0.90

 hurdle

0.89

 chunk

0.88

 misconception

0.87

 bang

0.87

 takeaway

0.85

 proponent

0.85

 prize

0.82

Activations Density 0.028%

references to superlatives or extremes

instances of the word "biggest" indicating significant or noteworthy events

No Comments

No Known Activations