INDEX

Explanations

phrases related to geographical locations and cultural terms

oai_token-act-pair · gpt-3.5-turbo

specific cultural or geographic references, particularly names and places related to indigenous populations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 quick

-0.70

 financial

-0.69

 mainstream

-0.67

 predictable

-0.67

 warranted

-0.66

 dishonest

-0.66

 corporate

-0.65

 good

-0.65

 believable

-0.65

 business

-0.64

POSITIVE LOGITS

inav

1.35

ija

1.21

inka

1.21

aceae

1.21

anus

1.19

inus

1.17

inia

1.16

enza

1.15

avia

1.15

unda

1.15

Activations Density 0.618%

phrases related to geographical locations and cultural terms

specific cultural or geographic references, particularly names and places related to indigenous populations

No Comments

No Known Activations