INDEX

Explanations

duplicated letters in words

oai_token-act-pair · gpt-3.5-turbo

repeated letter combinations or expressions of laughter

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Cogn

-0.77

ipal

-0.76

 Malt

-0.74

 Lead

-0.71

 Prospect

-0.71

 Scand

-0.69

 Valve

-0.68

 Stack

-0.68

 Mesa

-0.68

 Polar

-0.67

POSITIVE LOGITS

HH

3.02

hhhh

2.68

hh

2.53

hhh

2.53

HHHH

2.34

aaaa

2.04

eeee

1.90

OOOO

1.81

aaa

1.80

ww

1.76

Activations Density 0.043%

duplicated letters in words

repeated letter combinations or expressions of laughter

No Comments

No Known Activations