INDEX

Explanations

phrases related to giving additional explanations or elaborating on a specific topic

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate uncertainty or ambiguity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 simulator

-0.64

lim

-0.49

 Bengal

-0.49

 suspended

-0.48

 Cull

-0.47

 accustomed

-0.47

Registered

-0.46

 decaying

-0.45

 pursu

-0.45

 mills

-0.45

POSITIVE LOGITS

>:

0.60

asures

0.60

llah

0.57

etheless

0.57

phabet

0.55

mberg

0.55

rue

0.54

details

0.54

rium

0.53

cific

0.52

Activations Density 1.364%

phrases related to giving additional explanations or elaborating on a specific topic

phrases that indicate uncertainty or ambiguity

No Comments

No Known Activations