* GradientAI Auto ROPE Base calculation
https://gradient.ai/blog/scaling-rotational-embeddings-for-long-context-language-models
has a formula that better fits the ideal rope scaling.
Tested with Lllama3, checked calculation is correct for llama2. Retains logic for not scaling rope if under trained CTX.
* add in solar scaling logic
Solar based models require the context values to be multiplied by 8. This is (i'm guessing) because the positions as based on a 32k context, but sliding window of 4k.
* Update model_adapter.h
adding in tensor count to identify solar models based on tensor count of 435.
* Update model_adapter.cpp
add in n_tensor count for solar identification
* refactor and cleanup GradientAI rope scaling
---------
Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com>
* Update Vulkan RoPE implementation
* Return nullptr on alloc_buffer when allocation fails, instead of throwing an exception
Minor fixes
* Fix segfault when running out of VRAM
Co-authored-by: slaren <slarengh@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
* try to fix CUDA ci with --allow-unsupported-compiler
* trigger when build.yml changes
* another test
* try exllama/bdashore3 method
* install vs build tools before cuda toolkit
* try win-2019
This commit adds pull_request_template.md and CONTRIBUTING.md . It focuses on explaining to contributors the need to rate PR complexity level, when to add [no ci] and how to format PR title and descriptions.
Co-authored-by: Brian <mofosyne@gmail.com>
Co-authored-by: compilade <git@compilade.net>
This allows local build options (like LLAMA_*) to be set in the local
file instead of having to edit Makefile, or provide a long gmake command
line on every build.
Using '-include' avoids generating a warning if Makefile.local doesn't
exist.
This avoids compile-time warnings with clang:
./expose.h:66:15: note: const member 'seed' will never be initialized
66 | const int seed;
| ^
No functional change intended.
In #7075, to fix the conversion of (some) models using model-00001-of-00001.safetensors instead of model.safetensors for a single model part we simply used the same logic as the part count to get the part names.
But this doesn't always work correctly, like when unusual additional model files like consolidated.safetensors in https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3 are present.
This commit matching both the prefix and the suffix of the model part names should fix this problem without breaking any previously-supported upstream models. But according to report by @teleprint-me there is still some
persistent problem, but shall do in the meantime.
Main changes of this PR is to consolidate GGUFWriter.add_key and GGUFWriter.add_val into GGUFWriter.add_key_value.
In addition use_temp_file is now opt-in instead of opt-out defaulting to False.
Also GGUFWriter now does not require output file name until when actually writing to it.
And GGUFWriter doesn't really need to eagerly prepare the data layout of the metadata
* server : Smart selection of available slot using Longest Common Substring
* add usage
* remove trailing whitespaces
* Use Longest Common Prefix (LCP) instead of LCS
* Rename argument
* vulkan : reuse parent extra for views
* Fix validation error when multiple compute contexts are used in a graph
---------
Co-authored-by: 0cc4m <picard12@live.de>
* avoid to get prompt in infill mode and embedding mode
* remove embedding mode
* refactor format
---------
Co-authored-by: wudexiang <wudexiang@bytedance.com>
Concedo
askmyteapot
0cc4m
Deven Mistry
Brian
Johannes Gäßler
Georgi Gerganov
slaren
Olivier Chafik
Jared Van Bortel
Ben Ashbaugh
Nicolás Pérez
mgroeber9110
Lexi
sasha0552
compilade
Christian Zhou-Zheng
intelmatt
woodx
pengxin99
Clint Herron
Mattheus Chediak