INDEX

Explanations

dates written in ordinal form (e.g., 1st, 21st)

oai_token-act-pair · gpt-3.5-turbo

occurrences of ordinal indicators related to rankings or placements

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

 Tycoon

-0.64

hare

-0.61

ORE

-0.61

 aside

-0.58

 figure

-0.58

 harmless

-0.56

gob

-0.56

 Galile

-0.56

#$

-0.53

 Gadget

-0.52

POSITIVE LOGITS

st

3.96

sts

1.93

sta

1.88

ST

1.78

stal

1.75

sth

1.62

sten

1.61

stic

1.57

stan

1.48

stre

1.48

Activations Density 0.029%

dates written in ordinal form (e.g., 1st, 21st)

occurrences of ordinal indicators related to rankings or placements

No Comments

No Known Activations