INDEX

Explanations

quotes or statements being read aloud

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reads" 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

»

-0.60

ether

-0.59

rir

-0.57

edge

-0.55

ollo

-0.55

 greater

-0.55

wage

-0.55

venge

-0.55

restricted

-0.54

san

-0.54

POSITIVE LOGITS

 reads

3.79

 read

2.25

reads

2.20

 reading

1.73

 writes

1.73

read

1.53

 readings

1.49

 listens

1.41

 Read

1.40

Reading

1.37

Activations Density 0.013%

quotes or statements being read aloud

instances of the word "reads" in various contexts

No Comments

No Known Activations