INDEX

Explanations

text related to writing and editing content

oai_token-act-pair · gpt-3.5-turbo

references to myths or misconceptions, particularly about sexual orientation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 reunited

-0.83

 culminating

-0.78

 unveiling

-0.77

 clad

-0.76

 glimps

-0.76

 inaugural

-0.74

 decorated

-0.74

 Elias

-0.73

 culminated

-0.73

 crowned

-0.72

POSITIVE LOGITS

Reason

1.18

Avoid

1.15

 NEVER

1.07

 Nope

1.06

unless

1.03

:(

1.03

Unless

1.03

 Avoid

1.02

 FALSE

1.01

 ALWAYS

1.00

Activations Density 0.812%

text related to writing and editing content

references to myths or misconceptions, particularly about sexual orientation

No Comments

No Known Activations