INDEX

Explanations

words related to something being difficult or challenging

oai_token-act-pair · gpt-3.5-turbo

the prefix "imp" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 dressing

-0.74

 forwards

-0.74

 brew

-0.72

 Brooke

-0.71

ND

-0.67

 Bryan

-0.67

 stored

-0.66

sat

-0.65

 broth

-0.65

 vetted

-0.64

POSITIVE LOGITS

imp

4.21

Imp

1.75

Imp

1.65

im

1.34

imb

1.08

IMP

1.06

imm

1.04

ims

1.00

isp

0.99

imp

0.99

Activations Density 0.014%

words related to something being difficult or challenging

the prefix "imp" in various contexts

No Comments

No Known Activations