INDEX

Explanations

phrases related to assigning blame or fault

oai_token-act-pair · gpt-3.5-turbo

references to assigning blame in various contexts

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

chan

-0.78

forms

-0.75

marked

-0.75

intend

-0.74

irl

-0.72

fo

-0.72

mare

-0.70

mens

-0.69

miss

-0.67

mental

-0.65

POSITIVE LOGITS

 blame

0.92

 blaming

0.85

 forgiven

0.77

 blames

0.75

 squarely

0.65

 forgiveness

0.64

awaru

0.62

 blamed

0.61

oshop

0.60

fulness

0.60

Activations Density 0.016%

phrases related to assigning blame or fault

references to assigning blame in various contexts

No Comments

No Known Activations