INDEX

Explanations

names or abbreviation initials followed by a particular letter grade

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names and titles

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

umbn

-0.69

saf

-0.68

éĹĺ

-0.66

ModLoader

-0.65

oided

-0.64

caps

-0.63

thirds

-0.63

Fair

-0.62

availability

-0.60

Reviewer

-0.59

POSITIVE LOGITS

.?

0.89

.,

0.82

.:

0.73

./

0.72

.;

0.71

ullivan

0.67

 Armour

0.64

.,"

0.63

uce

0.63

#$

0.63

Activations Density 0.064%

names or abbreviation initials followed by a particular letter grade

proper nouns, specifically names and titles

No Comments

No Known Activations