INDEX

Explanations

names containing "uh"

oai_token-act-pair · gpt-3.5-turbo

utterances of hesitation or filler phrases

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

BOOK

-0.79

 Painter

-0.67

å¥³

-0.67

BALL

-0.67

chnology

-0.67

 Reborn

-0.66

0000000000000000

-0.64

 shards

-0.63

OGR

-0.62

 thinner

-0.61

POSITIVE LOGITS

ahah

1.10

ansen

1.00

awk

0.98

undai

0.93

awks

0.92

annah

0.88

uge

0.88

ospital

0.85

arsh

0.83

hhhh

0.82

Activations Density 0.012%

names containing "uh"

utterances of hesitation or filler phrases

No Comments

No Known Activations