INDEX

Explanations

phrases related to loss or defeat in a competitive context

oai_token-act-pair · gpt-3.5-turbo

references to sports losses or political defeats

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ridges

-0.64

monds

-0.60

 intakes

-0.60

fir

-0.59

 eyed

-0.58

 hubs

-0.58

rison

-0.58

FG

-0.58

Redditor

-0.57

 intake

-0.56

POSITIVE LOGITS

ingly

0.85

aciously

0.79

achi

0.78

agi

0.74

vich

0.74

ables

0.74

hard

0.73

nces

0.72

acious

0.72

bite

0.72

Activations Density 0.054%

phrases related to loss or defeat in a competitive context

references to sports losses or political defeats

No Comments

No Known Activations