INDEX

Explanations

adjectives or adverbs indicating a strong degree or intensity

oai_token-act-pair · gpt-3.5-turbo

intensifiers or qualifiers that suggest extremity or emphasis

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

bow

-0.98

sw

-0.88

res

-0.88

fl

-0.85

 rental

-0.85

arm

-0.82

exp

-0.82

 rating

-0.81

 score

-0.81

 written

-0.81

POSITIVE LOGITS

probably

1.95

possibly

1.92

almost

1.87

maybe

1.87

perhaps

1.86

sometimes

1.86

likely

1.81

nothing

1.80

always

1.79

someone

1.78

Activations Density 0.096%

adjectives or adverbs indicating a strong degree or intensity

intensifiers or qualifiers that suggest extremity or emphasis

No Comments

No Known Activations