INDEX

Explanations

if-statements suggesting potential outcomes based on certain conditions being met or facts being true

oai_token-act-pair · gpt-3.5-turbo

statements reflecting conditional scenarios and potential consequences

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

azo

-0.75

+++

-0.68

cember

-0.65

 Usually

-0.63

 Introduced

-0.62

"],

-0.62

swick

-0.62

URA

-0.62

LESS

-0.61

many

-0.61

POSITIVE LOGITS

 indeed

0.84

 anything

0.83

any

0.73

 extrap

0.68

 accurate

0.64

 prol

0.64

 Nost

0.63

 coincidence

0.63

 correct

0.63

 ever

0.61

Activations Density 0.144%

if-statements suggesting potential outcomes based on certain conditions being met or facts being true

statements reflecting conditional scenarios and potential consequences

No Comments

No Known Activations