INDEX

Explanations

words related to a specific named entity, likely 'Rotterdam'

oai_token-act-pair · gpt-3.5-turbo

proper nouns related to locations or events

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

ammy

-0.85

 Manor

-0.68

 Practice

-0.67

 CLSID

-0.63

 Earl

-0.63

 lung

-0.63

Tie

-0.63

 Dive

-0.61

 Gork

-0.61

 Pace

-0.59

POSITIVE LOGITS

sonian

0.78

brid

0.74

icles

0.73

othe

0.71

vantage

0.70

erred

0.69

ciation

0.69

dimension

0.68

ilee

0.67

tered

0.67

Activations Density 0.000%

words related to a specific named entity, likely 'Rotterdam'

proper nouns related to locations or events

No Comments

No Known Activations