INDEX

Explanations

phrases or patterns related to computer code, such as variable names, syntax, and function calls

oai_token-act-pair · gpt-3.5-turbo

ticks or special characters used for emphasis or quotation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

phrine

-0.74

 lifes

-0.72

 Mamm

-0.71

 dividing

-0.71

 condem

-0.68

 mills

-0.65

 division

-0.64

 unfocusedRange

-0.64

ocard

-0.61

ampton

-0.61

POSITIVE LOGITS

taboola

0.91

eer

0.89

daq

0.89

ansas

0.87

lein

0.82

seq

0.82

DAQ

0.81

POSE

0.77

natureconservancy

0.76

PLE

0.76

Activations Density 0.010%

phrases or patterns related to computer code, such as variable names, syntax, and function calls

ticks or special characters used for emphasis or quotation

No Comments

No Known Activations