| .. |
|
evals
|
Add MMLU eval to github
|
2024-05-20 10:20:53 -07:00 |
|
instantiations
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
activations.h
|
Simplify FFW by using MatMul_4x4_Batch_Add.
|
2024-07-16 09:41:23 -07:00 |
|
common.cc
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
common.h
|
Further 1.02x prefill speedup from batch 64->512
|
2024-07-15 07:26:10 -07:00 |
|
configs.h
|
Improve readability with RepeatedAttentionWindowSizes
|
2024-07-11 09:11:46 -07:00 |
|
gemma-inl.h
|
Simplify FFW by using MatMul_4x4_Batch_Add.
|
2024-07-16 09:41:23 -07:00 |
|
gemma.cc
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
gemma.h
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
kv_cache.cc
|
Prep for sharding gemma.cc: split into kv_cache, tokenizer.
|
2024-07-02 09:31:06 -07:00 |
|
kv_cache.h
|
Prep for sharding gemma.cc: split into kv_cache, tokenizer.
|
2024-07-02 09:31:06 -07:00 |
|
ops.h
|
SVE build fix: avoid capturing vectors directly.
|
2024-07-11 08:43:56 -07:00 |
|
ops_test.cc
|
Simplify matmul: only 2 overloads
|
2024-07-11 06:58:42 -07:00 |
|
run.cc
|
Lint fix - string append, remove stale TODO
|
2024-07-08 04:11:21 -07:00 |
|
tokenizer.cc
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
tokenizer.h
|
7x compile time speedup: shard gemma.cc
|
2024-07-03 06:35:04 -07:00 |
|
weights.cc
|
Further simplification to ForEachTensor, thanks I.K.
|
2024-06-17 07:12:26 -07:00 |
|
weights.h
|
Refactor configurables.
|
2024-07-10 21:30:58 -07:00 |
ff34370aac