INDEX

Explanations

mentions of actions related to reducing or trimming something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "cut" related to reductions in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

agher

-0.78

apo

-0.70

ESSION

-0.69

heid

-0.67

Registered

-0.67

Grey

-0.66

 Grace

-0.65

Ã±a

-0.64

 Asylum

-0.63

avery

-0.63

POSITIVE LOGITS

thro

1.04

lasses

1.02

scene

0.96

backs

0.96

lass

0.96

cutting

0.93

scenes

0.91

cuts

0.88

cut

0.87

downs

0.84

Activations Density 0.020%

mentions of actions related to reducing or trimming something

instances of the word "cut" related to reductions in various contexts

No Comments

No Known Activations