INDEX

Explanations

phrases related to the television network TLC

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific television channel, particularly TLC

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 territory

-0.74

 Kahn

-0.67

 Fant

-0.66

 odds

-0.66

 probability

-0.64

 strikes

-0.64

 stopping

-0.64

 Hacker

-0.64

 whist

-0.63

Od

-0.63

POSITIVE LOGITS

LC

4.45

lc

1.68

LC

1.43

LS

1.42

LM

1.39

NC

1.32

LV

1.29

LP

1.28

LD

1.25

LL

1.23

Activations Density 0.017%

phrases related to the television network TLC

mentions of a specific television channel, particularly TLC

No Comments

No Known Activations