INDEX

Explanations

words indicating common occurrences or general trends

oai_token-act-pair · gpt-3.5-turbo

instances of the word "typically" and its variations, indicating a focus on general patterns or norms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Keys

-0.70

planes

-0.68

adra

-0.67

Sel

-0.65

le

-0.65

Ole

-0.65

 Orchestra

-0.65

tein

-0.64

Kut

-0.64

amins

-0.62

POSITIVE LOGITS

entimes

0.94

 enjoyed

0.77

 categorized

0.77

 frowned

0.77

 characterized

0.74

ensical

0.72

 interacted

0.72

 staffed

0.72

 regarded

0.71

 comprised

0.71

Activations Density 0.015%

words indicating common occurrences or general trends

instances of the word "typically" and its variations, indicating a focus on general patterns or norms

No Comments

No Known Activations