INDEX

Explanations

numbers in contexts related to various topics and entities

oai_token-act-pair · gpt-3.5-turbo

numbers associated with quantities or ages

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

EStream

-0.64

feature

-0.64

reference

-0.64

 jerk

-0.62

ipment

-0.62

atem

-0.61

plom

-0.59

feed

-0.59

reply

-0.58

Reviewer

-0.57

POSITIVE LOGITS

th

1.16

teenth

1.02

 consecutive

0.97

 months

0.95

teen

0.94

 years

0.93

een

0.93

 percent

0.91

%"

0.91

 stitches

0.91

Activations Density 0.272%

numbers in contexts related to various topics and entities

numbers associated with quantities or ages

No Comments

No Known Activations