INDEX

Explanations

phrases related to observations and explanations

oai_token-act-pair · gpt-3.5-turbo

verbs indicating observation, acknowledgment, or explanation of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

angular

-0.64

Holy

-0.62

 cruc

-0.62

ãĥ¯

-0.62

onut

-0.61

productive

-0.61

respective

-0.60

inal

-0.58

frac

-0.57

gins

-0.56

POSITIVE LOGITS

hirt

0.66

 nostalg

0.66

 Quote

0.65

creen

0.64

 Pitch

0.62

HAEL

0.62

]:

0.62

inger

0.62

:]

0.61

Cth

0.61

Activations Density 0.240%

phrases related to observations and explanations

verbs indicating observation, acknowledgment, or explanation of information

No Comments

No Known Activations