INDEX

Explanations

phrases related to returning or reflecting on past events

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

Data Type

torch.float32

Hook Point

blocks.8.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

orthodox

-0.77

imon

-0.71

icons

-0.67

shown

-0.67

ruff

-0.67

olor

-0.66

stant

-0.65

orst

-0.65

ording

-0.65

tops

-0.64

POSITIVE LOGITS

 undone

1.11

 forth

0.96

 ashore

0.92

 hither

0.88

 into

0.81

 apart

0.81

 roaring

0.81

 closer

0.80

leon

0.78

out

0.77

Activations Density 1.061%

phrases related to returning or reflecting on past events

No Comments

No Known Activations