INDEX

Explanations

proper nouns or titles, particularly those in uppercase

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "THE" in various contexts

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

fitting

-0.72

cd

-0.71

java

-0.71

uid

-0.70

let

-0.69

hm

-0.68

iod

-0.67

fired

-0.67

ander

-0.66

perties

-0.66

POSITIVE LOGITS

ORY

1.15

 LAST

1.15

 FORM

1.10

 IMAGES

1.09

 STORY

1.05

ATER

1.05

IVERS

1.03

ISM

1.03

 WEEK

1.02

 DARK

1.02

Activations Density 0.019%

proper nouns or titles, particularly those in uppercase

occurrences of the word "THE" in various contexts

No Comments

No Known Activations