Skip to content

Actions: cchuter/llama.cpp

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
65 workflow runs
65 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
Eval bug: feat/v4-port-cuda AI review (issues) #3: Issue #8 opened by Talle
1d 0h 0m 1s
1d 0h 0m 1s
Benchmarking on sm86 machine with DDR4 offloading AI review (issues) #2: Issue #7 opened by kevinyu14
1d 0h 0m 1s
1d 0h 0m 1s
ROCm/HIP compatibility: managed memory fallback, FP8 path, UMA memory calc Pull Request Labeler #8: Pull request #6 opened by JDL440
35s
35s
ggml: add GGML_SCHED_MAX_SPLIT_INPUTS compile def Pull Request Labeler #7: Pull request #5 opened by Gustav-Simonsson
18s
18s
Eval bug: After adding the --parallel parameter, it cannot run properly AI review (issues) #1: Issue #4 opened by zt1024
1d 0h 0m 2s
1d 0h 0m 2s
llama: add DeepSeek V4 Flash model + architecture (CPU) Python check requirements.txt #8: Commit d99045f pushed by cchuter
5m 17s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
5m 17s
llama: add DeepSeek V4 Flash model + architecture (CPU) Check Pre-Tokenizer Hashes #8: Commit d99045f pushed by cchuter
1m 2s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
1m 2s
llama: add DeepSeek V4 Flash model + architecture (CPU) Python Type-Check #11: Commit d99045f pushed by cchuter
3m 24s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
3m 24s
llama: add DeepSeek V4 Flash model + architecture (CPU) Python Type-Check #10: Commit d99045f pushed by cchuter
3m 53s review/dsv4-cpu-ops-model
review/dsv4-cpu-ops-model
3m 53s
Merge origin/master into v4-clean-base (+20 upstream commits) Python Type-Check #9: Commit c054032 pushed by cchuter
3m 6s v4-clean-base
v4-clean-base
3m 6s
Merge origin/master into v4-clean-base (+20 upstream commits) Check Pre-Tokenizer Hashes #7: Commit c054032 pushed by cchuter
56s v4-clean-base
v4-clean-base
56s
Merge origin/master into v4-clean-base (+20 upstream commits) Python check requirements.txt #7: Commit c054032 pushed by cchuter
5m 22s v4-clean-base
v4-clean-base
5m 22s
llama: DeepSeek V4 Flash architecture (CPU, DSA-free) Python Type-Check #8: Commit 1da43b2 pushed by cchuter
2m 46s v4-clean-base
v4-clean-base
2m 46s
Python check requirements.txt Python check requirements.txt #6: Commit f979cbe pushed by cchuter
7m 2s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
7m 2s
Python Type-Check Python Type-Check #7: Commit f979cbe pushed by cchuter
3m 14s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
3m 14s
Check Pre-Tokenizer Hashes Check Pre-Tokenizer Hashes #6: Commit f979cbe pushed by cchuter
1m 6s upstream-pr/dsv4-ops-cpu
upstream-pr/dsv4-ops-cpu
1m 6s
cuda: software-FP16 fallback for fattn dequantize_V<half> on CC < 600 Pull Request Labeler #6: Pull request #3 opened by cchuter
22s
22s
cuda: software-FP16 fallback for fattn dequantize_V<half> on CC < 600 CI #6: Pull request #3 opened by cchuter
1d 0h 0m 3s fix/fattn-half-dequantize
fix/fattn-half-dequantize
1d 0h 0m 3s
cuda: software-FP16 fallback for fattn dequantize_V<half> on CC < 600 HIP quality check #6: Pull request #3 opened by cchuter
5m 24s fix/fattn-half-dequantize
fix/fattn-half-dequantize
5m 24s
cuda: software-FP16 fallback for fattn dequantize_V<half> on CC < 600 CI (self-hosted) #6: Pull request #3 opened by cchuter
1d 0h 0m 3s fix/fattn-half-dequantize
fix/fattn-half-dequantize
1d 0h 0m 3s
cuda: --split-mode row support for DeepSeek V4 (multi-GPU) CI #5: Pull request #2 synchronize by cchuter
1d 0h 0m 13s fix/v4-rowsplit
fix/v4-rowsplit
1d 0h 0m 13s
cuda: --split-mode row support for DeepSeek V4 (multi-GPU) CI (self-hosted) #5: Pull request #2 synchronize by cchuter
1d 0h 0m 2s fix/v4-rowsplit
fix/v4-rowsplit
1d 0h 0m 2s
cuda: --split-mode row support for DeepSeek V4 (multi-GPU) HIP quality check #5: Pull request #2 synchronize by cchuter
5m 12s fix/v4-rowsplit
fix/v4-rowsplit
5m 12s
cuda: --split-mode row support for DeepSeek V4 (multi-GPU) Server #5: Pull request #2 synchronize by cchuter
14m 38s fix/v4-rowsplit
fix/v4-rowsplit
14m 38s
cuda: --split-mode row support for DeepSeek V4 (multi-GPU) Pull Request Labeler #5: Pull request #2 synchronize by cchuter
16s
16s