INDEX

Explanations

terms related to locations or specific places

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or institutions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 protective

-0.70

itably

-0.67

ĻĤ

-0.67

 prone

-0.65

 oxide

-0.64

 verification

-0.63

 emergencies

-0.62

 pree

-0.61

 territorial

-0.61

 consumers

-0.60

POSITIVE LOGITS

 Room

1.16

 Awards

1.02

 Desk

1.01

 Graveyard

1.01

 Boys

0.99

 Guys

0.99

 Trilogy

0.98

 Programme

0.98

 Shop

0.98

 Bowl

0.98

Activations Density 0.317%

terms related to locations or specific places

references to specific locations or institutions

No Comments

No Known Activations