INDEX

Explanations

instances of specific words like 'beth', 'eth', 'oven', 'aul', and 'ier'

oai_token-act-pair · gpt-3.5-turbo

mentions of entities or names, particularly those related to classical composers

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

 slot

-0.70

 tremend

-0.70

 border

-0.66

 Malays

-0.62

 propensity

-0.62

 detail

-0.60

ashtra

-0.59

 frontier

-0.59

 easing

-0.56

 resultant

-0.56

POSITIVE LOGITS

zeb

0.98

lehem

0.88

leck

0.82

mingham

0.80

iful

0.79

reau

0.76

oven

0.75

nown

0.73

fleet

0.72

onge

0.71

Activations Density 0.054%

instances of specific words like 'beth', 'eth', 'oven', 'aul', and 'ier'

mentions of entities or names, particularly those related to classical composers

No Comments

No Known Activations