INDEX

Explanations

phrases related to issues or actions involving workers

oai_token-act-pair · gpt-3.5-turbo

mentions of "worker" or related terms indicating labor-related contexts

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

laus

-0.84

 venerable

-0.74

raints

-0.74

 Drawn

-0.68

olic

-0.67

Viz

-0.67

rawdownloadcloneembedreportprint

-0.67

nu

-0.64

aber

-0.64

 Liberties

-0.63

POSITIVE LOGITS

worker

1.17

 worker

1.13

 Worker

1.05

geist

0.84

workers

0.80

 cooper

0.77

handler

0.76

 workers

0.76

 bees

0.75

 ants

0.74

Activations Density 0.014%

phrases related to issues or actions involving workers

mentions of "worker" or related terms indicating labor-related contexts

No Comments

No Known Activations