INDEX

Explanations

specific instructions or lists

oai_token-act-pair · gpt-3.5-turbo

phrases or terms indicating lists, instructions, or details about sequential information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

)</

-0.77

DERR

-0.72

aukee

-0.72

±

-0.71

ipers

-0.70

Downloadha

-0.68

¶æ

-0.68

gran

-0.66

emy

-0.66

big

-0.62

POSITIVE LOGITS

:(

0.88

:-

0.80

 configure

0.71

 assumes

0.69

>:

0.68

:#

0.67

*:

0.67

viz

0.66

":[

0.63

Activations Density 0.094%

specific instructions or lists

phrases or terms indicating lists, instructions, or details about sequential information

No Comments

No Known Activations