ag-15/ashard_dry_run_transcript.txt at master · igor-holt/ag-15 · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
============================================================
aSHARD Runner - DRY-RUN MODE
============================================================

CONFIGURATION:
  Model path: /tmp/simulated_model.gguf
  Container image: ghcr.io/ggerganov/llama.cpp:latest
  GPU layers: 18
  Context size: 8192
  Threads: 4
  Network mode: default

PLANNED COMMANDS (would execute if not --dry-run):

[1/4] PODMAN PULL:
  podman pull ghcr.io/ggerganov/llama.cpp:latest

[2/4] GPU DETECTION:
  nvidia-smi --query-gpu=gpu_name,memory.total --format=csv

[3/4] PODMAN RUN:
  podman run  --device nvidia.com/gpu=all \
    -v /tmp/vibe-scratchpad-d5f27282-cq85hrkr/a2a/hermes-openclaw-single-pass:/workspace \
    -v /tmp/simulated_model.gguf:/model.gguf \
    -e MODEL_PATH=/model.gguf \
    -e GPU_LAYERS=18 \
    -e CTX_SIZE=8192 \
    -e THREADS=4 \
    --rm \
    ghcr.io/ggerganov/llama.cpp:latest \
    llama-cli \
      --model /model.gguf \
      --n-gpu-layers 18 \
      --ctx-size 8192 \
      --threads 4 \
      --prompt "[aSHARD] System prompt loaded\nUser: Hello" \
      --tokens 128 \
      --log-disabled

[4/4] HERMES AGENT:
  hermes-agent \
    --model /model.gguf \
    --ctx-size 8192 \
    --gpu-layers 18

============================================================
DRY-RUN COMPLETE: No commands were executed
Configuration is valid. To run for real, remove --dry-run
============================================================