MirrorHub/construct
0
0
Fork 0
mirror of https://github.com/matrix-construct/construct synced 2024-11-25 00:02:34 +01:00
Code Issues Releases Wiki Activity

ircd::gpt::pipe: Enable mutable model; fixes for backpropagation; range stub.

Browse source
This commit is contained in:
Jason Volk 2022-10-16 21:16:55 +00:00
parent 47117dde9a
commit 2609c21913
4 changed files with 15 additions and 13 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
include/ircd/gpt/model.h
View file

@ -30,6 +30,7 @@ namespace ircd::gpt::model
extern std::vector<json::object> default_data;
constexpr auto alignment {4096};
extern conf::item<bool> cache_shared;
}
/// Layer normalization

13
ircd/gpt.cc
View file

@ -256,10 +256,15 @@ try
}
,model
{
std::make_unique<pipe::model>
(
*const_cast<const gpt::model::decoder *>(gpt::model::default_model)
)
!gpt::model::cache_shared?
std::make_unique<pipe::model>
(
*const_cast<const gpt::model::decoder *>(gpt::model::default_model)
):
std::make_unique<pipe::model>
(
*const_cast<gpt::model::decoder *>(gpt::model::default_model)
)
}
,desc
{

12
ircd/gpt_gpu.cl
View file

@ -1167,7 +1167,7 @@ ircd_gpt_prop_elem(__global const struct ircd_gpt_ctrl *const ctrl,
const float4
param = param_[li],
grad = ctrl->label[0].loss.mean,
grad = ctrl->target.loss.mean,
alpha[2] = { 1.0f - opts->beta[0], 1.0f - opts->beta[1], },
exp_avg = ts? exp_avg_[li]: 0.0f,
exp_avg_sqr = ts? exp_avg_sqr_[li]: 0.0f,
@ -1179,13 +1179,9 @@ ircd_gpt_prop_elem(__global const struct ircd_gpt_ctrl *const ctrl,
delta = opts->alpha * (exp_avg_dot / denom),
update = param - delta;
param_[li] = param + FLT_EPSILON;
exp_avg_[li] = exp_avg + FLT_EPSILON;
exp_avg_sqr_[li] = exp_avg_sqr + FLT_EPSILON;
//param_[li] = update;
//exp_avg_[li] = exp_avg_dot;
//exp_avg_sqr_[li] = exp_avg_sqr_dot;
param_[li] = update;
exp_avg_[li] = exp_avg_dot;
exp_avg_sqr_[li] = exp_avg_sqr_dot;
}
//

2
ircd/gpt_pipe.cc
View file

@ -238,7 +238,7 @@ ircd::gpt::pipe::cycle::cycle(gpt::samp &samp)
tokens,
cached,
true,
false,
((false) && gpt::model::cache_shared)
}
,stage
{