INDEX

Explanations

incidents or situations where someone or a group is left in a negative or vulnerable state

oai_token-act-pair · gpt-3.5-turbo

discussions centered around vulnerability and the impact on individuals or groups

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

wcsstore

-0.70

imum

-0.64

oho

-0.63

pai

-0.61

 Excellence

-0.61

ATIONAL

-0.60

haps

-0.60

thren

-0.58

 Towards

-0.58

 Invasion

-0.58

POSITIVE LOGITS

 vulnerable

1.25

 scrambling

1.23

 stranded

1.22

 powerless

1.19

 baffled

1.16

 unable

1.15

 reeling

1.10

 susceptible

1.10

 unsatisf

1.09

 dissatisfied

1.09

Activations Density 0.099%

incidents or situations where someone or a group is left in a negative or vulnerable state

discussions centered around vulnerability and the impact on individuals or groups

No Comments

No Known Activations