INDEX

Explanations

dates in the format "month" + day, ignoring the year

oai_token-act-pair · gpt-3.5-turbo

the word "une" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

İĭ

-0.84

loo

-0.76

onial

-0.75

draw

-0.73

ories

-0.73

etheless

-0.72

pread

-0.71

aceous

-0.70

lished

-0.67

rieg

-0.67

POSITIVE LOGITS

arthed

1.04

quist

0.72

cker

0.70

cean

0.68

hill

0.65

SPONSORED

0.63

 buggy

0.63

Mik

0.63

chan

0.61

hall

0.59

Activations Density 0.020%

dates in the format "month" + day, ignoring the year

the word "une" in various contexts

No Comments

No Known Activations