INDEX

Explanations

years related to the late 1990s

oai_token-act-pair · gpt-3.5-turbo

references to the year 1997 and related events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

hop

-0.69

holder

-0.69

edo

-0.69

holders

-0.67

oÄŁ

-0.67

hatt

-0.66

tro

-0.65

ysc

-0.64

Tu

-0.63

 thirds

-0.63

POSITIVE LOGITS

ĸļ

0.94

-'

0.80

ãĥĥãĥī

0.76

ãĤ¦ãĤ¹

0.75

å¹

0.72

 onward

0.69

ãĥ¼ãĥ«

0.67

ãĥŁ

0.65

 onwards

0.65

âķĲ

0.62

Activations Density 0.020%

years related to the late 1990s

references to the year 1997 and related events

No Comments

No Known Activations