INDEX

Explanations

words related to analysis or evaluation, particularly expressions of assessment and measurement

oai_token-act-pair · gpt-3.5-turbo

statements expressing uncertainty or qualifications about performance or outcomes

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 decency

-0.64

lad

-0.61

fuck

-0.59

abouts

-0.58

 Pars

-0.57

rather

-0.57

tar

-0.57

ogi

-0.56

ows

-0.55

esses

-0.55

POSITIVE LOGITS

 necessarily

1.22

 exhaustive

1.18

 infall

1.09

 perfect

1.07

 uniformly

1.05

 universally

1.05

 exclusively

1.01

perfect

0.99

completely

0.98

 absolute

0.96

Activations Density 0.553%

words related to analysis or evaluation, particularly expressions of assessment and measurement

statements expressing uncertainty or qualifications about performance or outcomes

No Comments

No Known Activations