INDEX

Explanations

instances of the word "well" used as an interjection or indicating agreement

oai_token-act-pair · gpt-3.5-turbo

the word "Well" with varying frequencies of activation

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

amas

-0.67

 dash

-0.66

anca

-0.64

aca

-0.62

acters

-0.61

agne

-0.61

 punishable

-0.59

 identifying

-0.59

ibi

-0.58

 replacing

-0.56

POSITIVE LOGITS

 Well

3.38

Well

2.59

well

1.82

 well

1.44

Hmm

1.35

 Alright

1.35

 Anyway

1.34

 Yeah

1.33

 Turns

1.23

 Okay

1.22

Activations Density 0.012%

instances of the word "well" used as an interjection or indicating agreement

the word "Well" with varying frequencies of activation

No Comments

No Known Activations