INDEX

Explanations

terms related to resistance or protection against potential harm or damage

oai_token-act-pair · gpt-3.5-turbo

terms related to safety, resilience, and health conditions

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

initions

-0.70

ructure

-0.66

ocation

-0.65

ocations

-0.65

 Observatory

-0.63

comings

-0.63

rikes

-0.63

iage

-0.62

 Amendments

-0.61

 Orchestra

-0.60

POSITIVE LOGITS

worthy

1.27

 friendly

1.16

able

1.13

compatible

1.13

 oriented

1.06

 dependent

1.05

proof

1.05

 resistant

1.04

 sensitive

1.03

less

1.03

Activations Density 0.503%

terms related to resistance or protection against potential harm or damage

terms related to safety, resilience, and health conditions

No Comments

No Known Activations