Skip to content

Actions: beehive-lab/GPULlama3.java

Actions

Rerun Workflows

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
rerun-workflow.yml
71 workflow runs
71 workflow runs

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
OpenAI-compatible server (llama-tornado --server): /v1/chat/completions + /v1/completions, streaming, zero-dep Rerun Workflows #71: Issue comment #135 (comment) created by mikepapadim
11s
11s
On-device greedy sampling (-Dllama.deviceSample): GPU argmax, keep logits device-side (~500x less D2H) Rerun Workflows #70: Issue comment #134 (comment) created by mikepapadim
1s
1s
Add llama-bench-style benchmark: llama-tornado --bench (pp/tg/pg matrix, avg±stddev t/s, md/csv/json) Rerun Workflows #69: Issue comment #133 (comment) created by mikepapadim
1s
1s
Add llama-bench-style benchmark: llama-tornado --bench (pp/tg/pg matrix, avg±stddev t/s, md/csv/json) Rerun Workflows #68: Issue comment #133 (comment) created by mikepapadim
1s
1s
Add llama-bench-style benchmark: llama-tornado --bench (pp/tg/pg matrix, avg±stddev t/s, md/csv/json) Rerun Workflows #67: Issue comment #133 (comment) created by mikepapadim
Skipped
Skipped
Roadmap: vLLM-style serving features for TornadoVM batched decode Rerun Workflows #66: Issue comment #130 (comment) created by mikepapadim
Skipped
Skipped
Static batched decode (LLaMA + Qwen3): B independent sequences per step, up to 41x aggregate throughput Rerun Workflows #65: Issue comment #129 (comment) created by mikepapadim
1s
1s
Static batched decode (LLaMA + Qwen3): B independent sequences per step, up to 41x aggregate throughput Rerun Workflows #64: Issue comment #129 (comment) created by mikepapadim
Skipped
Skipped
Static batched decode (LLaMA + Qwen3): B independent sequences per step, up to 41x aggregate throughput Rerun Workflows #63: Issue comment #129 (comment) created by mikepapadim
1s
1s
Release 0.5.0 Rerun Workflows #62: Issue comment #126 (comment) created by CLAassistant
8s
8s
Release 0.5.0 Rerun Workflows #61: Issue comment #125 (comment) created by CLAassistant
8s
8s
Add pure CUDA backend along the PTX engine Rerun Workflows #60: Issue comment #124 (comment) created by orionpapadakis
1s
1s
Add pure CUDA backend along the PTX engine Rerun Workflows #59: Issue comment #124 (comment) created by mikepapadim
1s
1s
Add pure CUDA backend along the PTX engine Rerun Workflows #58: Issue comment #124 (comment) created by mikepapadim
7s
7s
Add pure CUDA backend along the PTX engine Rerun Workflows #57: Issue comment #124 (comment) created by mikepapadim
7s
7s
Add pure CUDA backend along the PTX engine Rerun Workflows #56: Issue comment #124 (comment) created by mikepapadim
1s
1s
Gemma 4 model support (CPU + GPU/TornadoVM, BF16 and Q8_0) Rerun Workflows #55: Issue comment #120 (comment) created by stratika
1s
1s
Add Q4_K/Q5_K/Q6_K GPU support via Q8_0 dequantization Rerun Workflows #54: Issue comment #108 (comment) created by orionpapadakis
7s
7s
Improve performance metrics collection Rerun Workflows #53: Issue comment #104 (comment) created by orionpapadakis
1s
1s
Rest API support Rerun Workflows #52: Issue comment #49 (comment) created by orionpapadakis
2s
2s
RMS normalization kernel optimization by fusing the reduction kernel and context mapping kernel Rerun Workflows #51: Issue comment #77 (comment) created by stratika
2s
2s
[WIP][models] Add support for Google's Gemma3 models Rerun Workflows #50: Issue comment #61 (comment) created by stratika
1s
1s
Several fixes and improvements for CI Rerun Workflows #49: Issue comment #115 (comment) created by stratika
1s
1s
Add Prefill–Decode Separation with Batched Prompt Ingestion and Logits Skipping Rerun Workflows #48: Issue comment #102 (comment) created by orionpapadakis
1s
1s
Add Q4_K/Q5_K/Q6_K GPU support via Q8_0 dequantization Rerun Workflows #47: Issue comment #108 (comment) created by CLAassistant
1s
1s