diff options
Diffstat (limited to 'examples/train-text-from-scratch')
| -rw-r--r-- | examples/train-text-from-scratch/CMakeLists.txt | 1 | ||||
| -rw-r--r-- | examples/train-text-from-scratch/train-text-from-scratch.cpp | 38 |
2 files changed, 21 insertions, 18 deletions
diff --git a/examples/train-text-from-scratch/CMakeLists.txt b/examples/train-text-from-scratch/CMakeLists.txt index 1a44c49..4459516 100644 --- a/examples/train-text-from-scratch/CMakeLists.txt +++ b/examples/train-text-from-scratch/CMakeLists.txt @@ -1,4 +1,5 @@ set(TARGET train-text-from-scratch) add_executable(${TARGET} train-text-from-scratch.cpp) +install(TARGETS ${TARGET} RUNTIME) target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT}) target_compile_features(${TARGET} PRIVATE cxx_std_11) diff --git a/examples/train-text-from-scratch/train-text-from-scratch.cpp b/examples/train-text-from-scratch/train-text-from-scratch.cpp index afbb4a7..54dc2be 100644 --- a/examples/train-text-from-scratch/train-text-from-scratch.cpp +++ b/examples/train-text-from-scratch/train-text-from-scratch.cpp @@ -16,6 +16,8 @@ #pragma warning(disable: 4244 4267) // possible loss of data #endif +static const float rms_norm_eps = LLAMA_DEFAULT_RMS_EPS; + struct random_normal_distribution { std::mt19937 gen; std::normal_distribution<float> rd; @@ -439,7 +441,7 @@ struct ggml_tensor * forward( // norm { // cur shape [n_embd,N,1,1] - cur = ggml_rms_norm(ctx0, inpL); + cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps); // cur = attention_norm*cur cur = ggml_mul(ctx0, @@ -562,7 +564,7 @@ struct ggml_tensor * forward( // norm { // cur shape [n_embd,N,1,1] - cur = ggml_rms_norm(ctx0, inpFF); + cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps); // cur = ffn_norm*cur // cur shape [n_embd,N,1,1] @@ -606,7 +608,7 @@ struct ggml_tensor * forward( { // inpL shape [n_embd,N,1,1] - inpL = ggml_rms_norm(ctx0, inpL); + inpL = ggml_rms_norm(ctx0, inpL, rms_norm_eps); // inpL = norm*inpL // inpL shape [n_embd,N,1,1] @@ -694,7 +696,7 @@ struct ggml_tensor * forward_batch( // norm { // cur shape [n_embd,N*n_batch,1,1] - cur = ggml_rms_norm(ctx0, inpL); + cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = attention_norm*cur @@ -857,7 +859,7 @@ struct ggml_tensor * forward_batch( // norm { // cur shape [n_embd,N*n_batch,1,1] - cur = ggml_rms_norm(ctx0, inpFF); + cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = ffn_norm*cur @@ -910,7 +912,7 @@ struct ggml_tensor * forward_batch( { // inpL shape [n_embd,N*n_batch,1,1] - inpL = ggml_rms_norm(ctx0, inpL); + inpL = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(inpL, n_embd, N*n_batch); // inpL = norm*inpL @@ -979,7 +981,7 @@ struct ggml_tensor * forward_batch_wo_cache( // norm { // cur shape [n_embd,N*n_batch,1,1] - cur = ggml_rms_norm(ctx0, inpL); + cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = attention_norm*cur @@ -1085,7 +1087,7 @@ struct ggml_tensor * forward_batch_wo_cache( // norm { // cur shape [n_embd,N*n_batch,1,1] - cur = ggml_rms_norm(ctx0, inpFF); + cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = ffn_norm*cur @@ -1138,7 +1140,7 @@ struct ggml_tensor * forward_batch_wo_cache( { // inpL shape [n_embd,N*n_batch,1,1] - inpL = ggml_rms_norm(ctx0, inpL); + inpL = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(inpL, n_embd, N*n_batch); // inpL = norm*inpL @@ -1203,7 +1205,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn( // norm { - cur = ggml_rms_norm(ctx0, inpL); + cur = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = attention_norm*cur @@ -1267,7 +1269,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn( { // norm { - cur = ggml_rms_norm(ctx0, inpFF); + cur = ggml_rms_norm(ctx0, inpFF, rms_norm_eps); assert_shape_2d(cur, n_embd, N*n_batch); // cur = ffn_norm*cur @@ -1311,7 +1313,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn( // norm { - inpL = ggml_rms_norm(ctx0, inpL); + inpL = ggml_rms_norm(ctx0, inpL, rms_norm_eps); assert_shape_2d(inpL, n_embd, N*n_batch); // inpL = norm*inpL @@ -1434,7 +1436,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn_train( gf->perf_time_us = 0; const auto & hparams = model->hparams; - //const int n_ctx = hparams.n_ctx; + const int n_ctx = hparams.n_ctx; const int n_vocab = hparams.n_vocab; const int n_embd = hparams.n_embd; const int n_layer = hparams.n_layer; @@ -1603,7 +1605,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn_train( struct my_llama_layer & layer = model->layers[il]; // tensors with values necessary for backward pass are in persistent buf(-1) // other tensors with buf(0) and buf(1) are only temporary needed, and their memory reused after layer is completed. - use_buf(-1); struct ggml_tensor * t02 = expand(gf, ggml_rms_norm (ctx0, cur)); assert_shape_2d(t02, n_embd, N*n_batch); + use_buf(-1); struct ggml_tensor * t02 = expand(gf, ggml_rms_norm (ctx0, cur, rms_norm_eps)); assert_shape_2d(t02, n_embd, N*n_batch); use_buf( 0); struct ggml_tensor * t03 = expand(gf, ggml_repeat (ctx0, layer.attention_norm, t02)); assert_shape_2d(t03, n_embd, N*n_batch); use_buf(-1); struct ggml_tensor * t04 = expand(gf, ggml_mul (ctx0, t02, t03)); assert_shape_2d(t04, n_embd, N*n_batch); use_buf(-1); struct ggml_tensor * t05 = expand(gf, ggml_mul_mat (ctx0, layer.wq, t04)); assert_shape_2d(t05, n_embd, N*n_batch); @@ -1623,7 +1625,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn_train( use_buf(-1); struct ggml_tensor * t19 = expand(gf, ggml_reshape_2d (ctx0, t18, n_embd, N*n_batch)); assert_shape_2d(t19, n_embd, N*n_batch); use_buf( 0); struct ggml_tensor * t20 = expand(gf, ggml_mul_mat (ctx0, layer.wo, t19)); assert_shape_2d(t20, n_embd, N*n_batch); use_buf(-1); struct ggml_tensor * t21 = expand(gf, ggml_add (ctx0, t20, cur)); assert_shape_2d(t21, n_embd, N*n_batch); - use_buf(-1); struct ggml_tensor * t22 = expand(gf, ggml_rms_norm (ctx0, t21)); assert_shape_2d(t22, n_embd, N*n_batch); + use_buf(-1); struct ggml_tensor * t22 = expand(gf, ggml_rms_norm (ctx0, t21, rms_norm_eps)); assert_shape_2d(t22, n_embd, N*n_batch); use_buf( 0); struct ggml_tensor * t23 = expand(gf, ggml_repeat (ctx0, layer.ffn_norm, t22)); assert_shape_2d(t23, n_embd, N*n_batch); use_buf(-1); struct ggml_tensor * t24 = expand(gf, ggml_mul (ctx0, t23, t22)); assert_shape_2d(t24, n_embd, N*n_batch); use_buf(-1); struct ggml_tensor * t25 = expand(gf, ggml_mul_mat (ctx0, layer.w3, t24)); assert_shape_2d(t25, n_ff, N*n_batch); @@ -1666,7 +1668,7 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn_train( } clr_buf(0); use_buf(0); - struct ggml_tensor * t31 = expand(gf, ggml_rms_norm (ctx0, cur)); assert_shape_2d(t31, n_embd, N*n_batch); + struct ggml_tensor * t31 = expand(gf, ggml_rms_norm (ctx0, cur, rms_norm_eps)); assert_shape_2d(t31, n_embd, N*n_batch); struct ggml_tensor * t32 = expand(gf, ggml_repeat (ctx0, model->norm, t31)); assert_shape_2d(t32, n_embd, N*n_batch); struct ggml_tensor * t33 = expand(gf, ggml_mul (ctx0, t32, t31)); assert_shape_2d(t33, n_embd, N*n_batch); use_buf(-1); @@ -1863,10 +1865,10 @@ struct ggml_tensor * forward_batch_wo_cache_flash_attn_train( t12->grad = expand(gb, ggml_permute(ctx0, t15->grad, 0, 2, 3, 1)); assert_shape_4d(t12->grad, N, n_batch, n_embd/n_head, n_head); t11->grad = expand(gb, ggml_reshape_2d(ctx0, ggml_cont(ctx0, t12->grad), N*n_batch, n_embd)); assert_shape_2d(t11->grad, N*n_batch, n_embd); t10->grad = expand(gb, ggml_permute(ctx0, t14->grad, 0, 2, 1, 3)); assert_shape_4d(t10->grad, n_embd/n_head, n_head, N, n_batch); - t09->grad = expand(gb, ggml_rope_back(ctx0, t10->grad, n_past, n_rot, rope_mode)); assert_shape_4d(t09->grad, n_embd/n_head, n_head, N, n_batch); + t09->grad = expand(gb, ggml_rope_back(ctx0, t10->grad, n_past, n_rot, rope_mode, n_ctx)); assert_shape_4d(t09->grad, n_embd/n_head, n_head, N, n_batch); t08->grad = expand(gb, ggml_reshape_2d(ctx0, t09->grad, n_embd, N*n_batch)); assert_shape_2d(t08->grad, n_embd, N*n_batch); t07->grad = expand(gb, ggml_permute(ctx0, t13->grad, 0, 2, 1, 3)); assert_shape_4d(t07->grad, n_embd/n_head, n_head, N, n_batch); - t06->grad = expand(gb, ggml_rope_back(ctx0, t07->grad, n_past, n_rot, rope_mode)); assert_shape_4d(t06->grad, n_embd/n_head, n_head, N, n_batch); + t06->grad = expand(gb, ggml_rope_back(ctx0, t07->grad, n_past, n_rot, rope_mode, n_ctx)); assert_shape_4d(t06->grad, n_embd/n_head, n_head, N, n_batch); t05->grad = expand(gb, ggml_reshape_2d(ctx0, t06->grad, n_embd, N*n_batch)); assert_shape_2d(t05->grad, n_embd, N*n_batch); t04->grad = expand(gb, ggml_add_inplace(ctx0, ggml_add_inplace(ctx0, |