INDEX

Explanations

the phrase "did not" followed by a verb

oai_token-act-pair · gpt-3.5-turbo

negations or phrases indicating a lack of action or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

undrum

-0.68

acca

-0.68

FTWARE

-0.67

rish

-0.61

hitting

-0.60

APTER

-0.60

Imagine

-0.59

Goo

-0.59

ablishment

-0.58

advertisement

-0.57

POSITIVE LOGITS

 nonetheless

1.49

 nevertheless

1.33

 beware

0.97

etheless

0.94

 differs

0.93

 retains

0.92

 cautioned

0.91

 differed

0.90

 later

0.90

 still

0.89

Activations Density 0.433%

the phrase "did not" followed by a verb

negations or phrases indicating a lack of action or information

No Comments

No Known Activations