ddd2691551
fix muse glimmer assistant tag
2026-08-12 18:13:57 +08:00
b3d0475aae
muse glimmer templates
2026-08-10 21:44:56 +08:00
71801743c4
added support for cohere north mini
2026-06-16 00:24:30 +08:00
093067ec01
support apriel
2026-06-15 23:39:40 +08:00
ae60ea0009
handle updated gemma templates
2026-04-14 00:29:58 +08:00
a33eda3842
more template fixes for the gemma4 31b
2026-04-04 14:23:16 +08:00
e6a49c404f
fixed gemma4 templates, basic non thinking is working fine on all models now
2026-04-03 12:20:01 +08:00
34ad53e950
merged support for gemma4. the e2b, e4b and 26b work, the 31b does not
2026-04-03 11:07:46 +08:00
ad0618e351
bump defaults, updated lite, fixed glm4.7 autoguess template
2026-02-21 08:51:53 +08:00
a82c89b065
minimax template
2026-01-04 20:51:16 +08:00
5fe1d51c24
fix gpt oss
2025-11-26 23:44:56 +08:00
9919fdc83a
granite 4 template
2025-10-08 00:09:46 +08:00
8b5ebfcc8d
updated lite, fix kimi template (+2 squashed commit)
...
Squashed commit:
[a2df7f1bd] alternate kimi template
[8a312a9f1] updated lite
2025-09-25 20:44:09 +08:00
f9998b2a5d
seed OSS chat templates
2025-08-23 18:01:39 +08:00
258f53a938
autoguess: fix Kimi search strings ( #1681 )
...
* updated lite
* autoguess: fix Kimi search strings
* Update klite.embd
---------
Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com >
2025-08-11 21:55:42 +08:00
204739e7f1
Adapter fixes ( #1659 )
...
* test adapters
* add assistant_gen adapter key
* add support for chat templates stored as .jinja files
* removed mistakenly commited gated-tokenizers link
* autoguess: Harmony: add missing newline prefixes to system_end
2025-08-10 16:19:50 +08:00
866cc346ab
tweak OpenAI Harmony autoguess developer prefix and assistant end token ( #1673 )
...
* tweak OpenAI Harmony autoguess developer prefix
* use <|end|> for adapter end
2025-08-08 21:15:11 +08:00
338b1fe97e
readjusted mistral and oai template, fixed compile issue on termux, updated lite, show generated token ids in debug mode
2025-08-07 21:14:48 +08:00
34487d3c02
gpt oss harmony template
2025-08-06 11:39:40 +08:00
9fbbd9e127
half measure for mistral spaced formats
2025-08-04 23:48:11 +08:00
ba626b3f4c
fixes for known flaws in chat adapter. Trailing space changes in mistral not included
2025-07-27 17:00:26 +08:00
f25339c92b
handle empty objects returned by tool calls, also remove misinterpretation of the tools calls instruct tag within ChatML autoguess
2025-07-25 22:22:27 +08:00
ff8f156fa0
AutoGuess tests ( #1650 )
...
* whitespace
* AutoGuess remove dot suffix in names
* .gitignore update
* test: added autoguess test suite
* github workflow to run autoguess test when appropriate
* git clone unavailable tokenizer configs rather than committing to repo
* fix link to included tokenizer configs
* skip storing downloaded tokenizer configs
* typo
* minor fixes
* clean-up
* limit workflow to trigger from experimental branch
---------
Co-authored-by: Concedo <39025047+LostRuins@users.noreply.github.com >
2025-07-25 19:21:00 +08:00
e1849f9a0b
updated lite (+4 squashed commit)
...
Squashed commit:
[ecaa653df1154c44a7db2e5e43d490b13af8
2025-07-20 11:10:03 +08:00
226624639c
AutoGuess: Move Generic cases to end of file and put Kimi with other ChatML variants ( #1648 )
...
* AutoGuess: Move Generic cases to end of file and put Kimi with other ChatML variants
* patch Kimi ChatML template
2025-07-18 13:24:21 +08:00
8396add5be
removed hunyuan autoguess template, fixed multi file loading up to 999 parts
2025-07-15 17:49:49 +08:00
f3f6168f85
added new autoguess templates
2025-07-15 14:51:20 +08:00
b649f69e4f
gemma3n chat template
2025-06-29 15:24:55 +08:00
9c27ccde50
RWKV World chat adapters ( #1612 )
2025-06-20 21:39:59 +08:00
ae8f01c2d4
no need for chat start for now
2025-05-22 23:19:29 +08:00
8b6dfbd1be
disabling the gMask prefix for glm-4 completions
2025-05-21 17:29:24 +08:00
d4316aa4ed
fixed bad template
2025-05-15 00:44:48 +08:00
11984f1040
fixed autoguess adapters, fixed tool builds
2025-05-13 19:38:56 +08:00
fda682fa12
updated lite
2025-04-30 19:49:54 +08:00
4d8a7a6594
fix occasional clip segfault, fix glm4 (+1 squashed commits)
...
Squashed commits:
[bd71cd688] GLM4 fix wip
2025-04-29 01:42:50 +08:00
7cb815b727
AutoGuess: GLM-4 ( #1502 )
...
* AutoGuess: GLM-4
* add 'chat_start' field to adapters
* GLM-4 fix
2025-04-26 08:47:42 +08:00
d32d0b382a
glm4 template
2025-04-25 00:41:15 +08:00
a6149ad0fc
fixed g3 adapter back
2025-04-12 23:17:54 +08:00
c94aec1930
update workflows, update gemma default adapter sysprompt
2025-04-12 18:38:23 +08:00
efef14bb82
added llama4 tags
2025-04-12 08:58:04 +08:00
93a226d9e4
added prefix for llava, reverted system role in template as it degreaded gemma3. truncated debug logs
2025-04-05 18:06:41 +08:00
669311365c
fixed gemma system prompt
2025-04-02 13:58:51 +08:00
0ed95fcccc
fixed l3 template, add index
2025-03-31 23:59:06 +08:00
6a1dd57435
gemma3 template, updated lite, fixed tool calling, reenable ctx shift for gemma3
2025-03-14 17:47:01 +08:00
6b7d2349a7
Rewrite history to fix bad vulkan shader commits without increasing repo size
...
added dpe colab (+8 squashed commit)
Squashed commit:
[b8362da4] updated lite
[ed6c037d] move nsigma into the regular sampler stack
[ac5f61c6] relative filepath fixed
[05fe96ab] export template
[ed0a5a3e] nix_example.md: refactor (#1401 )
* nix_example.md: add override example
* nix_example.md: drop graphics example, already basic nixos knowledge
* nix_example.md: format
* nix_example.md: Vulkan is disabled on macOS
Disabled in: https://github.com/NixOS/nixpkgs/commit/1ccd253accfb140acbf9da9f56d9dc6d3aece088
* nix_examples.md: nixpkgs.config.cuda{Arches -> Capabilities}
Fixes: https://github.com/LostRuins/koboldcpp/issues/1367
[675c62f7] AutoGuess: Phi 4 (mini) (#1402 )
[4bf56982b8c0df04#1397 )
- place after nsigma and before xtc (+3 squashed commit)
Squashed commit:
[87c52b97ee8906f3e85c0e69#1394 )
* stop counting reps
* fix range-based initializer
* strike that - reverse it
2025-03-05 00:02:20 +08:00
1cb9805024
add autoguess adapter for DeepSeek V2.5/R1 ( #1329 )
2025-01-22 20:39:04 +08:00
23ec550835
PoC: add chat template heuristics ( #1283 )
...
* PoC: add chat template heuristics
The fallback chat template adapter of Vicuna is not ideal in some cases (e.g. a test against a sub-portion of the BBC news classification task on Kaggle gave an 82% accuracy with Vicuna and 88% with the official ChatML format for a q4_k_m Qwen 2.5 3B-Instruct gguf).
This PR adds a proof of concept simple heuristic which looks at the chat template and upgrades the adapter when it is able to.
* gemma 2 heuristic
* Phi 4, Llama 3.x heuristics
* better qwen vs generic heuristic
* cleanup
* mistral (generic) heuristic
* fix sys msg for mistral
* phi 3.5
* mistral v3
* cohere (aya expanse 32b based)
* only derive from chat template if AutoGuess
* add notes about alpaca fallbacks
* added AutoGuess.json dummy
* add mistral v7
* switch to using a json list with search strings
2024-12-28 12:15:23 +08:00
Concedo
kallewoof
henk717