INDEX

Explanations

phrases tagged with "/**9" which seem to be related to requests or urgent situations

oai_token-act-pair · gpt-3.5-turbo

occurrences of asterisks or bullet points in the text

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

ly

-0.81

liness

-0.75

 dispers

-0.69

liest

-0.67

ijk

-0.67

iop

-0.65

ilia

-0.65

ilic

-0.64

aea

-0.63

 relocation

-0.62

POSITIVE LOGITS

kw

1.03

hole

0.83

Edited

0.79

THIS

0.79

orks

0.78

Discussion

0.77

Madison

0.77

DOWN

0.75

ski

0.74

WARNING

0.74

Activations Density 0.016%

phrases tagged with "/**9" which seem to be related to requests or urgent situations

occurrences of asterisks or bullet points in the text

No Comments

No Known Activations