INDEX

Explanations

the word "full" in the text

oai_token-act-pair · gpt-3.5-turbo

references to "full" 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 @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 preced

-0.61

mone

-0.59

 Vital

-0.59

ubb

-0.59

 alone

-0.57

 Tycoon

-0.57

DERR

-0.57

pher

-0.57

%]

-0.55

atto

-0.54

POSITIVE LOGITS

erton

1.25

 blown

1.17

fledged

1.13

blown

1.06

 fled

1.06

screen

1.03

 complement

0.96

 spectrum

0.88

grown

0.87

blooded

0.87

Activations Density 0.041%

the word "full" in the text

references to "full" in various contexts

No Comments

No Known Activations