INDEX

Explanations

words enclosed in brackets followed by specific actions or consequences

oai_token-act-pair · gpt-3.5-turbo

instances of closing quotation marks

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

Ń·

-0.90

ĪĴ

-0.86

ãĥł

-0.74

apon

-0.69

ĸļ

-0.64

ãĤ©

-0.63

icter

-0.63

©¶æ

-0.62

azine

-0.62

ngth

-0.62

POSITIVE LOGITS

VEN

0.70

>]

0.68

TPS

0.67

Management

0.66

âĨ

0.66

egal

0.65

...]

0.65

*)

0.64

ccording

0.64

Recent

0.63

Activations Density 0.032%

words enclosed in brackets followed by specific actions or consequences

instances of closing quotation marks

No Comments

No Known Activations