INDEX

Explanations

words related to speech or speech-related activities

oai_token-act-pair · gpt-3.5-turbo

references to speech and communication

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

 Soldiers

-0.70

 Royale

-0.65

astern

-0.65

 Associ

-0.63

aro

-0.63

risome

-0.62

Ã±o

-0.62

ellen

-0.61

 Popular

-0.61

 Jaguar

-0.61

POSITIVE LOGITS

speech

0.95

 speech

0.94

 imped

0.91

writers

0.86

writer

0.86

writing

0.80

emic

0.79

anooga

0.79

 uttered

0.78

ible

0.78

Activations Density 0.023%

words related to speech or speech-related activities

references to speech and communication

No Comments

No Known Activations