INDEX

Explanations

specific years in the 1990s

oai_token-act-pair · gpt-3.5-turbo

references to specific years, particularly the 1980s and 1990s

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

semble

-0.72

Edge

-0.68

edge

-0.66

bow

-0.66

 lightsaber

-0.66

jriwal

-0.66

arent

-0.65

ndra

-0.65

holder

-0.61

hire

-0.61

POSITIVE LOGITS

ĸļ

0.97

ãĥŁ

0.78

å¹

0.75

-'

0.73

ãĥ©ãĥ³

0.72

ãĥł

0.70

ãĥĥãĥĪ

0.69

rade

0.68

 census

0.68

aji

0.67

Activations Density 0.044%

specific years in the 1990s

references to specific years, particularly the 1980s and 1990s

No Comments

No Known Activations