INDEX

Explanations

conditional statements starting with "Had"

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "Had" indicating hypothetical scenarios or conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 outp

-0.64

 neigh

-0.64

FTWARE

-0.62

 repay

-0.60

 juggling

-0.59

ilings

-0.59

 shake

-0.59

 shedding

-0.59

 dish

-0.59

 scrimmage

-0.58

POSITIVE LOGITS

iths

1.05

 been

1.00

luck

0.95

ith

0.93

kson

0.92

been

0.91

rons

0.89

hers

0.88

oop

0.86

ron

0.86

Activations Density 0.102%

conditional statements starting with "Had"

phrases that include the word "Had" indicating hypothetical scenarios or conditions

No Comments

No Known Activations