INDEX

Explanations

words that include the substring "ro" followed by a high activation token possibly associated with progress or development

oai_token-act-pair · gpt-3.5-turbo

the word "ro" in various contexts, likely indicating a focus on variations of the word or its different usages

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

ãĥ¼ãĥĨãĤ£

-0.88

 EntityItem

-0.81

éĹĺ

-0.73

 Norn

-0.71

htt

-0.70

ioxide

-0.66

 Sons

-0.66

 Yankee

-0.66

IELD

-0.65

CPC

-0.65

POSITIVE LOGITS

ving

1.08

aches

1.04

vers

0.99

crow

0.98

dy

0.97

roach

0.97

java

0.97

cks

0.95

gged

0.95

ach

0.93

Activations Density 0.010%

words that include the substring "ro" followed by a high activation token possibly associated with progress or development

the word "ro" in various contexts, likely indicating a focus on variations of the word or its different usages

No Comments

No Known Activations