INDEX

Explanations

words associated with instructions or directives

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "to."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

listed

-0.66

soDeliveryDate

-0.65

hops

-0.64

pointers

-0.61

afety

-0.60

hur

-0.59

croft

-0.57

Got

-0.56

resy

-0.56

need

-0.56

POSITIVE LOGITS

pload

1.05

 maximize

0.96

 avoid

0.96

 compensate

0.96

 minimize

0.93

 give

0.91

 eliminate

0.90

 keep

0.90

 satisfy

0.89

 fulfill

0.88

Activations Density 1.202%

words associated with instructions or directives

occurrences of the word "to."

No Comments

No Known Activations