INDEX

Explanations

the word "Sad" with varying degrees of emphasis

oai_token-act-pair · gpt-3.5-turbo

references to or mentions of the word "Sad."

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

RAFT

-0.74

YP

-0.70

æ©Ł

-0.69

ambers

-0.69

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.68

 Reloaded

-0.67

 FANT

-0.67

ardless

-0.65

å§«

-0.64

ORGE

-0.64

POSITIVE LOGITS

 Pupp

1.00

omas

0.90

ness

0.90

hya

0.86

rament

0.85

emic

0.85

eway

0.85

ifiers

0.83

idy

0.82

eways

0.82

Activations Density 0.015%

the word "Sad" with varying degrees of emphasis

references to or mentions of the word "Sad."

No Comments

No Known Activations