INDEX

Explanations

phrases expressing certainty or conviction

oai_token-act-pair · gpt-3.5-turbo

references to the speaker and their perspective

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

urate

-0.69

licted

-0.67

pload

-0.65

abama

-0.65

eport

-0.64

shore

-0.63

cific

-0.61

selage

-0.60

pole

-0.60

rien

-0.60

POSITIVE LOGITS

SQL

0.66

¥µ

0.66

 wisely

0.62

 damned

0.62

AGA

0.61

senal

0.58

 caveats

0.56

selves

0.56

Audio

0.56

 Exception

0.56

Activations Density 0.107%

phrases expressing certainty or conviction

references to the speaker and their perspective

No Comments

No Known Activations