INDEX

Explanations

the number "one" appearing in different contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "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 @ 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

cases

-1.16

projects

-1.04

lations

-0.93

values

-0.93

types

-0.92

ooks

-0.92

forms

-0.90

groups

-0.89

levels

-0.89

olds

-0.88

POSITIVE LOGITS

 heck

1.01

 month

1.01

 hour

0.97

 year

0.95

 week

0.92

 hell

0.91

 hundred

0.88

 inning

0.88

 minute

0.87

 hurdle

0.86

Activations Density 0.088%

the number "one" appearing in different contexts

instances of the word "one" in various contexts

No Comments

No Known Activations