* examples/finetune -opt SGD (stochastic gradient descent) memory opt
add unit tested GGML_OPT_OPTIMIZER_SGD to ggml - avoids allocating
m, v tensors.
support finetune.cpp arg -opt SGD (or sgd). (default adamw as before)
llama 3.2-1b-F32 result: observed 11gb gpu ram (41 sec/epoch)
when using SGD instead of 19gb (55 sec/epoch) using adamw.
(wikipedia 100 lines finetune)
(
using the same GPU memory, adamw can only do before OOM 512
batch/context, reaching:
train: [███████▉] data=0000140/0000140 loss=0.02575±0.00099 acc=99.52±0.03% t=00:00:47 ETA=00:00:00
val: [███████▉] data=0000008/0000008 loss=4.76565±0.28810 acc=41.46±0.77% t=00:00:00 ETA=00:00:00
SGD is superior, though it converges slower, with max before OOM 1728
batch/context (esp see the better validation perf):
train: [███████▉] data=0000039/0000039 loss=0.00371±0.00010 acc=99.96±0.01% t=00:00:41 ETA=00:00:00
val: [███████▉] data=0000003/0000003 loss=5.11406±0.76034 acc=48.01±0.69% t=00:00:01 ETA=00:00:00
)
note: when finetuning long enough (or w/ enough -lr),
validation accuracy *eventually* drops ('catastrophic forgetting')
-lr-half (halflife) option useful for SGD to avoid oscillation or
super slow underdamped learning (makes setting -lr more forgiving).
terminal -lr for now is set by lr-halvings i.e. if you want at most
1/8 the inital -lr you set -lr-halvings 3.
note: objective loss not directly comparable between adamw, sgd? -
check perplexity or accuracy or consider relative improvements
for convergence
new finetune args -wd 1e-9 to enable weight decay in sgd or adamw,
and max -epochs N (default 2 as before)
cache (1 - wd*alpha) in 'adamw' opt struct -
no noticeable perf benefit, disabled (still done
for new SGD though)
since opt. memory is pre-allocated, the ggml_opt_get_optimizer_params
would probably be able to change between SGD and AdamW with each epoch
but would need to use adamw for the first (unconfirmed - no cmdline arg
to set such a policy yet)
test-opt checks adamw as before and now sgd (except for a few disabled
tests for sgd only; probably just needs logging values and adding
alternate reference values); tolerance on the 'regression'
test is broader for sgd (so we don't need many more epochs)
* Vulkan: Implement GGML_OP_OPT_STEP_SGD
* tests: Fix OPT_STEP_SGD test-backend-ops
* SGD op param store weight-decay and not 1-alpha*wd
* minor + cosmetic changes
* fix vulkan sgd
* try CI fix
---------
Co-authored-by: 0cc4m <picard12@live.de>
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs
Gemma3n uses Matrix-Matrix addition as part of their input processing,
wrongly triggering CUDA_GRAPH disablement on NVGPUs even when batch-size
of 1 is used.
* Exclude `project_per_layer_input` by matching node names
This ensures that all other graphs which don't exhibit this pattern do
not have their behavior changed.
* Revert unnecessary formatting changes
* CUDA: add set rows for f32 and f16
* Review: change kernel params, use strides from host
* Use 1-d kernel
* Review: use int64_t for blockDim.x, rename nb->s for clarity
* llama : initial Mamba-2 support
* ggml : SIMD ggml_ssm_scan for Mamba-2
* ggml : improve ggml_mul speed when masking recurrent states
* llama : support running Mamba-Codestral-7B-v0.1
* llama : fix Mamba-2 conv state saving
* ggml : make the ggml_mul fast broadcast path more consistently formatted
* llama : remove unused variable
* llama : add missing break
* convert_hf : prefer SentencePiece tokenizer for Mamba-2 when present
The tokenzier.json of Mamba-Codestral-7B-v0.1 otherwise requires
workarounds to work correctly.
* llama : avoid redundant state copy for Mamba 1 and 2
* metal : attempt to adapt SSM_SCAN for Mamba-2
* metal : fix SSM_SCAN pipeline scope
* metal : use log and exp instead of log1pf and expf in SSM_SCAN
* metal : remove unused arguments for SSM_SCAN
The max index is 31, so trimming the arguments is necessary.
* metal : add back n_seqs to SSM_SCAN args
Whoops, this is needed for the offset in the concatenated output.
* metal : fix SSM_SCAN state head offset
* metal : fix wrong number of tokens per sequence in SSM_SCAN
* ggml : remove unused fast broadcast path in GGML_MUL
This was initially added because states were masked with ggml_mul,
but this is no longer done and so this "optimisation" is no longer
necessary, or at least not worth the additional code complexity.
* ggml : avoid multiply by D in GGML_OP_SSM_SCAN
This makes the weight buft detection in src/llama.cpp simpler.
* convert : transpose Mamba-2 A, D and reshape SSM_NORM
This breaks existing conversions of Mamba-2 models
to avoid some reshapes.
Not sure if it's a good idea,
but it makes the graph slightly cleaner.
* llama : more appropriate SSM_SCAN and SSM_CONV buft support checks
* convert : fix flake8 lint
* metal : fix confusion between ; and ,
* metal : add missing args for nb references in ssm_scan_f32_group
* metal : single-user mamba2 inference works
* kv-cache : remove const_cast when setting inputs for s_copy
And also fix multi-user inference for recurrent models
by using cell_id instead of i as the kv cell index
when populating s_copy.
* convert : avoid AutoConfig for Mamba and Mamba2 hparams
* kv-cache : allow context shift for recurrent models
* graph : fix recurrent state copies when avoiding copies
Works, but using lambda functions might not be that clean.
* ggml : fix mamba2 ssm scan when compiled with SVE
* ggml-cpu : reorder SVE FMA for consistency with other SIMD arches
* cuda : implement ssm scan for Mamba2
There is still room for improvement, but it works!
* cuda : adapt Mamba1 ssm scan to shape changes from Mamba2
* mamba : fix mismatched new and delete size for llm_build_mamba
Subclasses of llm_graph_context cannot have extra fields,
because the called destructor is not the one from the subclass.
This otherwise would cause problems when runnning Mamba-(1|2) inference
when compiled -DGGML_SANITIZE_ADDRESS=ON
* cuda : graceful fallback for Mamba-1 models with weird embd size
* CUDA: add softmax broadcast
* Pass by const ref
* Review: Use blockDims for indexing, remove designated initializers
* Add TODO for noncontigous input/output
* implement unary REGLU/GEGLU/SWIGLU cpu ops
* relax constraints
* duplicate shape of source
* fix ggml_vec_geglu_f16
* special case gated ops
* implement unary REGLU/GEGLU/SWIGLU cuda ops
* tighten constraints again
* refactor into GGML_GLU_OP
* metal : add glu kernels
ggml-ci
* add CUDA_GLU_BLOCK_SIZE [no ci]
* more constraints and use 64bit ints
ggml-ci
* 64bit multiplication [no ci]
* implement swapped variants (cpu/cuda)
* update comment [no ci]
ggml-ci
* Vulkan: Add GLU ops and shaders
* SYCL: Implement fused kernel GEGLU, SWIGLU and REGLU for single up+gate
* ggml : implement GLU for split up/gate (#14181)
* implement GLU for split up/gate
* add tests for ggml_glu_split
* Vulkan: Implement glu_split logic and shader support
* add split to logging [no ci]
* SYCL: refactor element_size ops and add split up and gate support to gated kernels
* SYCL: switch GEGLU to use tanh approximation
---------
Co-authored-by: 0cc4m <picard12@live.de>
Co-authored-by: Akarshan <akarshan@menlo.ai>
* GGML: increase OP count in assertion
* Refactor: Optimize SYCL element-wise operations with unary function inlining
This commit refactors the SYCL element-wise operations to improve performance by:
- Inlining unary operations (sgn, abs, elu, gelu, silu, etc.) to reduce kernel launch overhead.
- Introducing helper functions `op_xxx` for each unary operation to encapsulate the logic.
- Replacing direct kernel calls with calls to these inlined functions.
- Using `__dpct_inline__` to encourage compiler inlining.
- Minor code cleanup and consistency improvements.
The changes aim to reduce kernel launch overhead and improve the overall efficiency of element-wise operations on SYCL devices.
* vulkan: Increase workgroup size for GLU, for performance (#14345)
* vulkan: Increase workgroup size for GLU, for performance
* vulkan: change GLU shaders to do one element per invocation rather than one row per workgroup
* merge fix
* metal : add support for split and swap
ggml-ci
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: 0cc4m <picard12@live.de>
Co-authored-by: Akarshan <akarshan@menlo.ai>
Co-authored-by: Jeff Bolz <jbolz@nvidia.com>
* CUDA: add bf16 and f32 support to cublas_mul_mat_batched
* Review: add type traits and make function more generic
* Review: make check more explicit, add back comments, and fix formatting
* Review: fix formatting, remove useless type conversion, fix naming for bools
* 1. add "integrated" in ggml_cuda_device_info for distinguish whether it is Intergrate_gpu or discrete_gpu
2. Adjust the func:"ggml_backend_cuda_device_supports_buft" for this new feature
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Adjusted code indentation
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Fixed incorrect setting of variable types
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Adjusted the judgment logic
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* add a host_buft assert in case of integrated_cuda_device with func:'evaluate_and_capture_cuda_graph()'
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Add a defensive security assert
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Adjusted the support judgment logic.
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* revoke the suggest commit changes due to it's not applicable in jetson_device
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Add parentheses to enforce operator precedence
Co-authored-by: Diego Devesa <slarengh@gmail.com>
* Update ggml/src/ggml-cuda/ggml-cuda.cu
Fix ci bug: add a spaces
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
---------
Co-authored-by: yangxiao <yang_xl@tju.edu.cn>
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
Co-authored-by: yangxiao <yangxl_zz@qq.com>
Co-authored-by: Diego Devesa <slarengh@gmail.com>
