INDEX

Explanations

phrases related to success or achievement

oai_token-act-pair · gpt-3.5-turbo

references to "victory."

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

ridges

-0.76

asus

-0.74

effic

-0.71

 pores

-0.69

minecraft

-0.65

 notor

-0.64

Factor

-0.61

 needles

-0.60

pper

-0.59

 compuls

-0.58

POSITIVE LOGITS

antly

0.86

hower

0.74

cry

0.72

lap

0.72

 victory

0.72

ceed

0.70

stroke

0.70

alist

0.70

bringer

0.70

ANCE

0.69

Activations Density 0.042%

phrases related to success or achievement

references to "victory."

No Comments

No Known Activations