INDEX

Explanations

references to "TT" which seems to be a specific term or acronym that holds significant importance

oai_token-act-pair · gpt-3.5-turbo

references to a specific trading or investment partnership

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

 Marilyn

-0.72

 Mesa

-0.67

 primates

-0.65

 incor

-0.63

 exponent

-0.63

 pepper

-0.63

 Emer

-0.63

 Gloria

-0.62

 lumin

-0.60

 Romero

-0.59

POSITIVE LOGITS

TT

4.43

TT

2.17

TG

1.53

tt

1.50

TPP

1.46

GG

1.34

TD

1.33

Ts

1.30

VT

1.26

TA

1.25

Activations Density 0.020%

references to "TT" which seems to be a specific term or acronym that holds significant importance

references to a specific trading or investment partnership

No Comments

No Known Activations