INDEX

Explanations

versions of software programs

oai_token-act-pair · gpt-3.5-turbo

periods or version numbers associated with software

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

 purse

-0.79

 affili

-0.74

 burglary

-0.71

 crocod

-0.71

 worsh

-0.68

 volunt

-0.67

amy

-0.67

 omission

-0.67

 hypnot

-0.66

 arms

-0.65

POSITIVE LOGITS

1.05

1.01

0.99

txt

0.94

jpg

0.93

exe

0.93

XXX

0.92

0.90

jar

0.89

0.88

Activations Density 0.108%

versions of software programs

periods or version numbers associated with software

No Comments

No Known Activations