INDEX

Explanations

occurrences of specific letters or patterns involving them

oai_token-act-pair · gpt-3.5-turbo

the presence of words starting with the letter 'd' in varying frequencies

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

éĹĺ

-0.85

Nun

-0.81

EStream

-0.78

 Remem

-0.77

 constitu

-0.77

å§«

-0.72

OIL

-0.69

 Forth

-0.66

 unanim

-0.65

 suscept

-0.64

POSITIVE LOGITS

ynamic

1.30

orm

1.22

iamond

1.21

etermin

1.19

etermined

1.16

rown

1.16

ownt

1.16

elta

1.16

ynam

1.14

iscover

1.13

Activations Density 0.031%

occurrences of specific letters or patterns involving them

the presence of words starting with the letter 'd' in varying frequencies

No Comments

No Known Activations