INDEX

Explanations

phrases related to additional information or content

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "more"

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

xtap

-0.77

IPS

-0.74

IP

-0.72

ress

-0.68

 Reconstruction

-0.68

west

-0.67

asar

-0.66

orescent

-0.66

ustomed

-0.66

ģ«

-0.64

POSITIVE LOGITS

 than

1.05

 Than

0.89

ado

0.89

 info

0.89

 expensive

0.81

 importantly

0.80

natureconservancy

0.78

 complicated

0.75

HUD

0.75

 stringent

0.75

Activations Density 0.089%

phrases related to additional information or content

occurrences of the word "more"

No Comments

No Known Activations