INDEX

Explanations

the word "negate" and variations of the word with different suffixes

oai_token-act-pair · gpt-3.5-turbo

terms related to negation or denial

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

favorite

-0.80

tale

-0.74

RAW

-0.73

lake

-0.68

 Gilmore

-0.68

chart

-0.66

eric

-0.66

 Hastings

-0.65

advertisement

-0.65

Hub

-0.64

POSITIVE LOGITS

neg

3.72

 negate

1.80

neg

1.17

mit

1.16

Neg

1.15

 fals

1.14

 negatives

1.14

 necess

1.13

 deval

1.12

Neg

1.08

Activations Density 0.026%

the word "negate" and variations of the word with different suffixes

terms related to negation or denial

No Comments

No Known Activations