INDEX

Explanations

words related to proper nouns, specifically names or titles that contain 'ald'

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals' last names

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

PER

-0.87

PLAY

-0.74

puter

-0.72

zzo

-0.72

senal

-0.71

PLA

-0.70

ORTS

-0.69

prime

-0.68

PROV

-0.67

kick

-0.67

POSITIVE LOGITS

sburg

1.09

orf

0.98

rums

0.95

ry

0.90

ivia

0.88

emort

0.88

oran

0.86

sson

0.85

ald

0.85

trump

0.85

Activations Density 0.015%

words related to proper nouns, specifically names or titles that contain 'ald'

references to specific individuals' last names

No Comments

No Known Activations