happyz
/
gemma.cpp
mirror of https://github.com/google/gemma.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[WIP] quality tweaks - for constants, defer float cast and use double for intermediate computations, add `model` to EOT token

This commit is contained in:
austinvhuang 2024-03-06 15:34:11 -05:00
parent 5b9d8a9936
commit 0f6a4b49d5
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
gemma.cc
View File

@ -295,7 +295,7 @@ HWY_NOINLINE void Attention(size_t batch_start, size_t batch_idx, size_t layer,
static constexpr size_t kModelDim =
gcpp::Activations<TConfig, kBatchSize>::kModelDim;
static constexpr size_t kHeads = TConfig::kHeads;
const float kQueryScale = 1.0 / sqrtf(static_cast<float>(kQKVDim));
static const float kQueryScale = static_cast<float>(1.0 / sqrt(static_cast<double>(kQKVDim)));
pool.Run(0, kHeads, [&](const uint64_t head, size_t /*thread*/) HWY_ATTR {
// linear projections to QKV
@ -418,7 +418,7 @@ HWY_NOINLINE void Prefill(const int* tokens, size_t num_tokens, size_t pos,
hwy::ThreadPool& inner_pool) {
PROFILER_ZONE("Gen.Prefill\\Att\\FFW");
static constexpr size_t kModelDim = TConfig::kModelDim;
static const float kEmbScaling = sqrtf(static_cast<float>(kModelDim));
static const float kEmbScaling = static_cast<float>(sqrt(static_cast<double>(kModelDim)));
pool.Run(
0, num_tokens, [&](const uint64_t token_idx, size_t /*thread*/) HWY_ATTR {
@ -473,7 +473,7 @@ void Transformer(int token, size_t pos,
static constexpr size_t kLayers = TConfig::kLayers;
static constexpr size_t kModelDim = TConfig::kModelDim;
static const float kEmbScaling = sqrtf(static_cast<float>(kModelDim));
static const float kEmbScaling = static_cast<float>(sqrt(static_cast<double>(kModelDim)));
Decompress(c_weights.c_embedder_input_embedding, token * kModelDim,
activations.x.data(), kModelDim);

2
run.cc
View File

@ -186,7 +186,7 @@ void ReplGemma(gcpp::Gemma& model, hwy::ThreadPool& pool,
if (abs_pos > 0) {
// Prepend "<end_of_turn>" token if this is a multi-turn dialogue
// continuation.
prompt_string = "<end_of_turn>\n" + prompt_string;
prompt_string = "<end_of_turn>model\n" + prompt_string;
}
}