INDEX

Explanations

mentions or references to TV shows, especially "The Daily Show"

oai_token-act-pair · gpt-3.5-turbo

mentions of television shows

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 membranes

-0.74

arta

-0.66

 cogn

-0.65

 liberties

-0.62

 agric

-0.61

Ì

-0.60

paralle

-0.60

 insofar

-0.60

RIC

-0.58

 solvent

-0.58

POSITIVE LOGITS

 Show

3.77

Show

2.43

 SHOW

2.27

show

2.27

 Shows

2.06

 show

1.96

shows

1.76

Hide

1.25

 shows

1.25

 Hide

1.23

Activations Density 0.015%

mentions or references to TV shows, especially "The Daily Show"

mentions of television shows

No Comments

No Known Activations