INDEX

Explanations

personal and location names, potentially from a transcription or dialogue

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly people's names and entities

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

 polluted

-0.71

 agric

-0.71

historic

-0.70

 centrally

-0.68

 institutional

-0.66

 polar

-0.65

 respectively

-0.65

BuyableInstoreAndOnline

-0.64

 centralized

-0.64

 consolidation

-0.64

POSITIVE LOGITS

laughter

1.04

laughs

1.01

aughed

0.86

ause

0.86

 laughs

0.85

Narr

0.85

 laughed

0.85

laugh

0.84

Pause

0.83

ridor

0.82

Activations Density 0.137%

personal and location names, potentially from a transcription or dialogue

proper nouns, particularly people's names and entities

No Comments

No Known Activations