INDEX

Explanations

phrases expressing approval or positive sentiment

oai_token-act-pair · gpt-3.5-turbo

positive expressions of achievement and well-being

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

urst

-0.77

umed

-0.75

¶ħ

-0.74

ume

-0.74

uming

-0.73

uria

-0.72

olina

-0.71

ipers

-0.71

agram

-0.71

cessive

-0.69

POSITIVE LOGITS

 comrade

0.79

tid

0.72

reen

0.72

clus

0.70

 folks

0.70

 inconvenience

0.69

 somebody

0.65

 avoids

0.65

noon

0.64

 sunshine

0.64

Activations Density 0.293%

phrases expressing approval or positive sentiment

positive expressions of achievement and well-being

No Comments

No Known Activations