INDEX

Explanations

the word "previously."

oai_token-act-pair · gpt-3.5-turbo

the word "previously" and its variations, indicating an emphasis on discussing prior states or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Cohn

-0.72

ĺħ

-0.68

owl

-0.64

ibility

-0.64

ility

-0.63

ĳå£«

-0.63

 Guys

-0.62

 centr

-0.61

enger

-0.60

rates

-0.60

POSITIVE LOGITS

 unseen

1.22

existing

1.21

 unsolved

1.08

mentioned

1.08

 unexpl

1.07

 unheard

1.07

 unre

1.07

 existing

1.06

 unex

1.05

 unused

1.05

Activations Density 0.072%

the word "previously."

the word "previously" and its variations, indicating an emphasis on discussing prior states or events

No Comments

No Known Activations