Fine-tuning LLMs to 1.58bit: extreme quantization made easy
•
148
Mohamed Mekkouri
medmekk
AI & ML interests
None yet
Articles
Organizations
models
12
medmekk/Meta-Llama-3-8B-torchao-int8_weight_only
Updated
•
11
medmekk/Llama-3.1-70B-torchao-int8_weight_only
Updated
•
42
medmekk/gpt2-torchao-int8_weight_only
Updated
•
8
medmekk/Meta-Llama-3-8B-torchao-int8_dynamic_activation_int8_weight
Updated
•
4
medmekk/Meta-Llama-3-8B-torchao-int4_weight_only-gs_256
Updated
•
4
medmekk/Meta-Llama-3-8B-torchao-int4_weight_only-gs-32
Updated
•
4
medmekk/Meta-Llama-3-8B-torchao-int4_weight_only-gs-64
Updated
•
4
medmekk/Meta-Llama-3-8B-quantized-int4_weight_only-2
Updated
•
4
medmekk/Meta-Llama-3-8B-quantized-int8_weight_only-2
Updated
•
4
medmekk/Meta-Llama-3-8B-quantized-int4_weight_only
Updated
•
4
datasets
None public yet