INDEX

Explanations

the word "one" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

references to something being described as "one" in various contexts

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

osponsors

-0.80

 actionGroup

-0.71

ories

-0.67

ooks

-0.66

ategory

-0.63

erity

-0.61

ivated

-0.61

avascript

-0.60

Available

-0.60

artments

-0.58

POSITIVE LOGITS

 Hundred

0.87

 hundred

0.85

 dimensional

0.77

 Thousand

0.74

esan

0.73

 thousand

0.69

stice

0.69

 sided

0.69

rency

0.66

advertisement

0.61

Activations Density 0.059%

the word "one" with a high level of activation

references to something being described as "one" in various contexts

No Comments

No Known Activations