INDEX

Explanations

words related to places or organizations

oai_token-act-pair · gpt-3.5-turbo

unique identifiers and mentions of specific events or entities

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

etheless

-0.77

 contested

-0.76

 tert

-0.72

perty

-0.72

 urgent

-0.70

 unmarried

-0.70

 carbohyd

-0.70

 challeng

-0.70

 permissible

-0.68

 nont

-0.67

POSITIVE LOGITS

 Wars

1.16

Nation

1.13

Gate

1.12

 Depot

1.12

 Monkey

1.11

 Nation

1.10

 Junction

1.07

 Squad

1.07

 Lounge

1.04

 Girl

1.04

Activations Density 0.340%

words related to places or organizations

unique identifiers and mentions of specific events or entities

No Comments

No Known Activations