INDEX

Explanations

dates and locations related to news events

oai_token-act-pair · gpt-3.5-turbo

dates and significant numerical information

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

dule

-0.66

course

-0.65

Reviewer

-0.62

 stages

-0.58

 Appropri

-0.58

 references

-0.57

ilty

-0.56

ometric

-0.56

 overw

-0.56

 overlap

-0.56

POSITIVE LOGITS

WASHINGTON

0.82

-+-+

0.79

FIELD

0.78

 YORK

0.75

esa

0.75

UES

0.74

EVA

0.74

POL

0.74

 CITY

0.73

APR

0.70

Activations Density 0.126%

dates and locations related to news events

dates and significant numerical information

No Comments

No Known Activations