INDEX

Explanations

keywords related to public speeches and addresses, especially with mention of specific days of the week

oai_token-act-pair · gpt-3.5-turbo

phrases related to speeches and addresses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 saline

-0.68

finder

-0.68

 intrusive

-0.61

rome

-0.59

 breeds

-0.59

 inhibitor

-0.58

IER

-0.58

ahn

-0.58

 allele

-0.58

 algae

-0.57

POSITIVE LOGITS

 speeches

1.14

 denouncing

0.93

writer

0.92

 delivered

0.90

 announcing

0.90

 speech

0.89

 Speech

0.88

writers

0.87

 lectures

0.82

 keynote

0.78

Activations Density 0.096%

keywords related to public speeches and addresses, especially with mention of specific days of the week

phrases related to speeches and addresses

No Comments

No Known Activations