INDEX

Explanations

instances where someone makes a comment or provides commentary

oai_token-act-pair · gpt-3.5-turbo

instances of the word "commented" or "remarked" indicating commentary or observations made by individuals

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

 hell

-0.69

 thieves

-0.64

cover

-0.61

 Freedom

-0.60

 robbers

-0.60

 heat

-0.59

 crisis

-0.58

ieve

-0.57

 rock

-0.56

 enemies

-0.55

POSITIVE LOGITS

 commented

3.66

 remarked

2.46

 commenting

1.68

 elaborated

1.54

 comment

1.50

 joked

1.49

 congratulated

1.49

 exclaimed

1.47

Comment

1.44

 complained

1.42

Activations Density 0.007%

instances where someone makes a comment or provides commentary

instances of the word "commented" or "remarked" indicating commentary or observations made by individuals

No Comments

No Known Activations