INDEX

Explanations

mentions of trends or patterns in data

oai_token-act-pair · gpt-3.5-turbo

instances of the word "trend" in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ricular

-0.79

gur

-0.75

 lungs

-0.69

idden

-0.69

feet

-0.67

hiro

-0.67

Lak

-0.66

ut

-0.65

porting

-0.64

Dam

-0.64

POSITIVE LOGITS

 trend

1.17

 Trend

1.05

setting

0.99

 trends

0.96

Trend

0.95

chart

0.84

LINE

0.81

 Trends

0.81

 reversal

0.79

 offender

0.79

Activations Density 0.005%

mentions of trends or patterns in data

instances of the word "trend" in the text

No Comments

No Known Activations