INDEX

Explanations

words related to telecommunication, specifically the word "telegram."

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "telegraph" and related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Augustine

-0.68

Tex

-0.67

Ware

-0.65

cca

-0.65

stay

-0.63

 Hispanics

-0.63

 Lank

-0.62

 Hispanic

-0.61

 Bain

-0.59

 vacation

-0.59

POSITIVE LOGITS

legraph

1.58

icz

0.92

kefeller

0.90

hedon

0.88

hester

0.88

ombat

0.87

 glim

0.84

icut

0.83

olson

0.80

anyahu

0.79

Activations Density 0.004%

words related to telecommunication, specifically the word "telegram."

occurrences of the word "telegraph" and related terms

No Comments

No Known Activations