INDEX

Explanations

terms related to a sequence of events happening in a continuous or uninterrupted manner

oai_token-act-pair · gpt-3.5-turbo

phrases indicating consecutive achievements or occurrences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

conom

-0.84

isk

-0.81

spoken

-0.78

OTO

-0.71

emis

-0.69

still

-0.68

XT

-0.67

detail

-0.66

hyde

-0.66

mere

-0.65

POSITIVE LOGITS

 seasons

0.90

 summers

0.87

 nights

0.86

 consecutive

0.86

 unsuccessful

0.84

 digits

0.83

 outings

0.82

 victories

0.82

 outing

0.82

 weekends

0.80

Activations Density 0.044%

terms related to a sequence of events happening in a continuous or uninterrupted manner

phrases indicating consecutive achievements or occurrences

No Comments

No Known Activations