INDEX

Explanations

mentions of attention or intention in the text

oai_token-act-pair · gpt-3.5-turbo

terms related to attention and intention

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Mehran

-0.78

senal

-0.75

accompan

-0.69

 crest

-0.68

 tribal

-0.66

 Juliet

-0.63

 Recon

-0.62

 shatter

-0.61

 Mend

-0.60

 succeeding

-0.60

POSITIVE LOGITS

ention

1.05

edly

0.90

aldehyde

0.88

rary

0.88

theless

0.86

ertodd

0.85

estinal

0.85

endment

0.83

rontal

0.83

ally

0.83

Activations Density 0.005%

mentions of attention or intention in the text

terms related to attention and intention

No Comments

No Known Activations