INDEX

Explanations

instances of the word "looked" with relatively high activation values

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "looked" in various contexts

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

ocon

-0.64

 indefinite

-0.58

 defect

-0.57

 passage

-0.57

 Free

-0.56

 Code

-0.56

 charter

-0.56

 donations

-0.55

 libel

-0.55

ensing

-0.55

POSITIVE LOGITS

 looked

3.29

 smelled

2.02

 looks

1.92

 glanced

1.92

 resembled

1.88

 tasted

1.88

 gazed

1.81

look

1.81

 look

1.79

 LOOK

1.75

Activations Density 0.012%

instances of the word "looked" with relatively high activation values

occurrences of the word "looked" in various contexts

No Comments

No Known Activations