Skip to content

finetune -ngl not offloading to GPU, Metal M2 #3911

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
QueryType opened this issue Nov 2, 2023 · 3 comments
Closed

finetune -ngl not offloading to GPU, Metal M2 #3911

QueryType opened this issue Nov 2, 2023 · 3 comments
Labels
bug-unconfirmed stale

Comments

@QueryType
Copy link

Prerequisites

Release: b1471

Expected Behavior

When I pass the command, with --gpu-layers 1, finetune should offload to GPU, .
/Volumes/d/apps/llama.cpp/llama.cpp/finetune --model-base /Volumes/d/apps/aimodels/others/openllama-3b-v2/openllama-3b-v2.q8_0.gguf --gpu-layers 1 --checkpoint-in chk-ol3b-shakespeare-LATEST.gguf --checkpoint-out chk-ol3b-shakespeare-ITERATION.gguf --lora-out lora-ol3b-shakespeare-ITERATION.bin --train-data shakespeare.txt --save-every 10 --threads 4 --adam-iter 30 --batch 4 --ctx 64 --use-checkpointing

Current Behavior

No, GPU used. All cores of P-CPU maxed out. Temperatures soaring. Check asitop, output.
Screenshot 2023-11-02 at 7 03 06 PM

Environment and Context

Mac Mini, M2, 24GB Memory

Failure Information (for bugs)

Please help provide information about the failure / bug.

Steps to Reproduce

  1. git checkout b1471
  2. make clean
  3. make
  4. Run the command for finetune as shared above, in current behavior
  5. Observe, asitop, or any other utility

Failure Logs

Volumes/d/apps/llama.cpp/llama.cpp/finetune  --model-base /Volumes/d/apps/aimodels/others/openllama-3b-v2/openllama-3b-v2.q8_0.gguf --gpu-layers 25 --checkpoint-in  chk-ol3b-shakespeare-LATEST.gguf --checkpoint-out chk-ol3b-shakespeare-ITERATION.gguf --lora-out lora-ol3b-shakespeare-ITERATION.bin --train-data shakespeare.txt --save-every 10  --threads 4 --adam-iter 30 --batch 4 --ctx 128 --use-checkpointing
main: seed: 1698931925
main: model base = '/Volumes/d/apps/aimodels/others/openllama-3b-v2/openllama-3b-v2.q8_0.gguf'
llama_model_loader: loaded meta data with 19 key-value pairs and 237 tensors from /Volumes/d/apps/aimodels/others/openllama-3b-v2/openllama-3b-v2.q8_0.gguf (version GGUF V3 (latest))
llama_model_loader: - tensor    0:                token_embd.weight q8_0     [  3200, 32000,     1,     1 ]
llama_model_loader: - tensor    1:              blk.0.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor    2:              blk.0.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor    3:              blk.0.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor    4:         blk.0.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor    5:            blk.0.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor    6:            blk.0.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor    7:              blk.0.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor    8:           blk.0.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor    9:            blk.0.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   10:              blk.1.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   11:              blk.1.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   12:              blk.1.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   13:         blk.1.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   14:            blk.1.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   15:            blk.1.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   16:              blk.1.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   17:           blk.1.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   18:            blk.1.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   19:              blk.2.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   20:              blk.2.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   21:              blk.2.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   22:         blk.2.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   23:            blk.2.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   24:            blk.2.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   25:              blk.2.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   26:           blk.2.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   27:            blk.2.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   28:              blk.3.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   29:              blk.3.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   30:              blk.3.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   31:         blk.3.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   32:            blk.3.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   33:            blk.3.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   34:              blk.3.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   35:           blk.3.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   36:            blk.3.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   37:              blk.4.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   38:              blk.4.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   39:              blk.4.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   40:         blk.4.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   41:            blk.4.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   42:            blk.4.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   43:              blk.4.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   44:           blk.4.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   45:            blk.4.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   46:              blk.5.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   47:              blk.5.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   48:              blk.5.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   49:         blk.5.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   50:            blk.5.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   51:            blk.5.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   52:              blk.5.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   53:           blk.5.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   54:            blk.5.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   55:              blk.6.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   56:              blk.6.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   57:              blk.6.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   58:         blk.6.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   59:            blk.6.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   60:            blk.6.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   61:              blk.6.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   62:           blk.6.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   63:            blk.6.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   64:              blk.7.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   65:              blk.7.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   66:              blk.7.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   67:         blk.7.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   68:            blk.7.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   69:            blk.7.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   70:              blk.7.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   71:           blk.7.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   72:            blk.7.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   73:              blk.8.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   74:              blk.8.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   75:              blk.8.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   76:         blk.8.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   77:            blk.8.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   78:            blk.8.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   79:              blk.8.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   80:           blk.8.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   81:            blk.8.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   82:              blk.9.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   83:              blk.9.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   84:              blk.9.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   85:         blk.9.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   86:            blk.9.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   87:            blk.9.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   88:              blk.9.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   89:           blk.9.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   90:            blk.9.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   91:             blk.10.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   92:             blk.10.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   93:             blk.10.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   94:        blk.10.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor   95:           blk.10.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   96:           blk.10.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor   97:             blk.10.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor   98:          blk.10.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor   99:           blk.10.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  100:             blk.11.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  101:             blk.11.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  102:             blk.11.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  103:        blk.11.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  104:           blk.11.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  105:           blk.11.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  106:             blk.11.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  107:          blk.11.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  108:           blk.11.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  109:             blk.12.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  110:             blk.12.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  111:             blk.12.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  112:        blk.12.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  113:           blk.12.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  114:           blk.12.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  115:             blk.12.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  116:          blk.12.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  117:           blk.12.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  118:             blk.13.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  119:             blk.13.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  120:             blk.13.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  121:        blk.13.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  122:           blk.13.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  123:           blk.13.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  124:             blk.13.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  125:          blk.13.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  126:           blk.13.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  127:             blk.14.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  128:             blk.14.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  129:             blk.14.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  130:        blk.14.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  131:           blk.14.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  132:           blk.14.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  133:             blk.14.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  134:          blk.14.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  135:           blk.14.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  136:             blk.15.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  137:             blk.15.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  138:             blk.15.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  139:        blk.15.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  140:           blk.15.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  141:           blk.15.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  142:             blk.15.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  143:          blk.15.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  144:           blk.15.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  145:             blk.16.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  146:             blk.16.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  147:             blk.16.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  148:        blk.16.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  149:           blk.16.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  150:           blk.16.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  151:             blk.16.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  152:          blk.16.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  153:           blk.16.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  154:             blk.17.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  155:             blk.17.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  156:             blk.17.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  157:        blk.17.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  158:           blk.17.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  159:           blk.17.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  160:             blk.17.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  161:          blk.17.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  162:           blk.17.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  163:             blk.18.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  164:             blk.18.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  165:             blk.18.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  166:        blk.18.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  167:           blk.18.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  168:           blk.18.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  169:             blk.18.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  170:          blk.18.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  171:           blk.18.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  172:             blk.19.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  173:             blk.19.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  174:             blk.19.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  175:        blk.19.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  176:           blk.19.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  177:           blk.19.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  178:             blk.19.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  179:          blk.19.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  180:           blk.19.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  181:             blk.20.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  182:             blk.20.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  183:             blk.20.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  184:        blk.20.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  185:           blk.20.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  186:           blk.20.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  187:             blk.20.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  188:          blk.20.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  189:           blk.20.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  190:             blk.21.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  191:             blk.21.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  192:             blk.21.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  193:        blk.21.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  194:           blk.21.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  195:           blk.21.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  196:             blk.21.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  197:          blk.21.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  198:           blk.21.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  199:             blk.22.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  200:             blk.22.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  201:             blk.22.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  202:        blk.22.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  203:           blk.22.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  204:           blk.22.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  205:             blk.22.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  206:          blk.22.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  207:           blk.22.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  208:             blk.23.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  209:             blk.23.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  210:             blk.23.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  211:        blk.23.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  212:           blk.23.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  213:           blk.23.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  214:             blk.23.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  215:          blk.23.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  216:           blk.23.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  217:             blk.24.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  218:             blk.24.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  219:             blk.24.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  220:        blk.24.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  221:           blk.24.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  222:           blk.24.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  223:             blk.24.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  224:          blk.24.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  225:           blk.24.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  226:             blk.25.attn_q.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  227:             blk.25.attn_k.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  228:             blk.25.attn_v.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  229:        blk.25.attn_output.weight q8_0     [  3200,  3200,     1,     1 ]
llama_model_loader: - tensor  230:           blk.25.ffn_gate.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  231:           blk.25.ffn_down.weight q8_0     [  8640,  3200,     1,     1 ]
llama_model_loader: - tensor  232:             blk.25.ffn_up.weight q8_0     [  3200,  8640,     1,     1 ]
llama_model_loader: - tensor  233:          blk.25.attn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  234:           blk.25.ffn_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  235:               output_norm.weight f32      [  3200,     1,     1,     1 ]
llama_model_loader: - tensor  236:                    output.weight q8_0     [  3200, 32000,     1,     1 ]
llama_model_loader: - kv   0:                       general.architecture str     
llama_model_loader: - kv   1:                               general.name str     
llama_model_loader: - kv   2:                       llama.context_length u32     
llama_model_loader: - kv   3:                     llama.embedding_length u32     
llama_model_loader: - kv   4:                          llama.block_count u32     
llama_model_loader: - kv   5:                  llama.feed_forward_length u32     
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32     
llama_model_loader: - kv   7:                 llama.attention.head_count u32     
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32     
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32     
llama_model_loader: - kv  10:                          general.file_type u32     
llama_model_loader: - kv  11:                       tokenizer.ggml.model str     
llama_model_loader: - kv  12:                      tokenizer.ggml.tokens arr     
llama_model_loader: - kv  13:                      tokenizer.ggml.scores arr     
llama_model_loader: - kv  14:                  tokenizer.ggml.token_type arr     
llama_model_loader: - kv  15:                tokenizer.ggml.bos_token_id u32     
llama_model_loader: - kv  16:                tokenizer.ggml.eos_token_id u32     
llama_model_loader: - kv  17:            tokenizer.ggml.padding_token_id u32     
llama_model_loader: - kv  18:               general.quantization_version u32     
llama_model_loader: - type  f32:   53 tensors
llama_model_loader: - type q8_0:  184 tensors
llm_load_vocab: special tokens definition check successful ( 259/32000 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 2048
llm_load_print_meta: n_embd           = 3200
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 32
llm_load_print_meta: n_layer          = 26
llm_load_print_meta: n_rot            = 100
llm_load_print_meta: n_gqa            = 1
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-06
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: n_ff             = 8640
llm_load_print_meta: freq_base_train  = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: model type       = 3B
llm_load_print_meta: model ftype      = mostly Q8_0
llm_load_print_meta: model params     = 3.43 B
llm_load_print_meta: model size       = 3.39 GiB (8.50 BPW) 
llm_load_print_meta: general.name   = others
llm_load_print_meta: BOS token = 1 '<s>'
llm_load_print_meta: EOS token = 2 '</s>'
llm_load_print_meta: UNK token = 0 '<unk>'
llm_load_print_meta: PAD token = 0 '<unk>'
llm_load_print_meta: LF token  = 13 '<0x0A>'
llm_load_tensors: ggml ctx size =    0.08 MB
llm_load_tensors: mem required  = 3472.53 MB
.................................................................................................
llama_new_context_with_model: n_ctx      = 512
llama_new_context_with_model: freq_base  = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_new_context_with_model: kv self size  =  162.50 MB
llama_build_graph: non-view tensors processed: 602/602
ggml_metal_init: allocating
ggml_metal_init: found device: Apple M2
ggml_metal_init: picking default device: Apple M2
ggml_metal_init: default.metallib not found, loading from source
ggml_metal_init: loading '/Volumes/d/apps/llama.cpp/llama.cpp/ggml-metal.metal'
ggml_metal_init: GPU name:   Apple M2
ggml_metal_init: GPU family: MTLGPUFamilyApple8 (1008)
ggml_metal_init: hasUnifiedMemory              = true
ggml_metal_init: recommendedMaxWorkingSetSize  = 16384.02 MB
ggml_metal_init: maxTransferRate               = built-in GPU
llama_new_context_with_model: compute buffer total size = 74.88 MB
llama_new_context_with_model: max tensor size =   103.76 MB
ggml_metal_add_buffer: allocated 'data            ' buffer, size =  3473.17 MB, ( 3473.80 / 16384.02)
ggml_metal_add_buffer: allocated 'kv              ' buffer, size =   162.52 MB, ( 3636.31 / 16384.02)
ggml_metal_add_buffer: allocated 'alloc           ' buffer, size =    68.77 MB, ( 3705.08 / 16384.02)
main: init model
print_params: n_vocab:   32000
print_params: n_ctx:     128
print_params: n_embd:    3200
print_params: n_ff:      8640
print_params: n_head:    32
print_params: n_head_kv: 32
print_params: n_layer:   26
print_params: norm_rms_eps          : 0.000001
print_params: rope_freq_base        : 10000.000000
print_params: rope_freq_scale       : 1.000000
print_lora_params: n_rank_attention_norm : 1
print_lora_params: n_rank_wq             : 4
print_lora_params: n_rank_wk             : 4
print_lora_params: n_rank_wv             : 4
print_lora_params: n_rank_wo             : 4
print_lora_params: n_rank_ffn_norm       : 1
print_lora_params: n_rank_w1             : 4
print_lora_params: n_rank_w2             : 4
print_lora_params: n_rank_w3             : 4
print_lora_params: n_rank_tok_embeddings : 4
print_lora_params: n_rank_norm           : 1
print_lora_params: n_rank_output         : 4
main: total train_iterations 0
main: seen train_samples     0
main: seen train_tokens      0
main: completed train_epochs 0
main: lora_size = 54798560 bytes (52.3 MB)
main: opt_size  = 81693904 bytes (77.9 MB)
main: opt iter 0
main: input_size = 65538080 bytes (62.5 MB)
main: compute_size = 5092979808 bytes (4857.0 MB)
main: evaluation order = RIGHT_TO_LEFT
main: tokenize training data
tokenize_file: total number of samples: 26702
main: number of training tokens: 26830
main: number of unique tokens: 3320
main: train data seems to have changed. restarting shuffled epoch.
main: begin training
main: work_size = 512240 bytes (0.5 MB)
train_opt_callback: iter=     0 sample=1/26702 sched=0.000000 loss=0.000000 |->
@tartakynov
Copy link

tartakynov commented Nov 22, 2023 •
edited
Loading

I think currently only f16 and f32 base models are supported for GPU offloading #3762 (comment)

@joaomdmoura
Copy link

Oh I hope they bring support for quantized models, it seems f16 don't work still actually

@github-actions github-actions bot added the stale label Mar 19, 2024
@github-actions GitHub Actions
Copy link
Contributor

github-actions bot commented Apr 2, 2024

This issue was closed because it has been inactive for 14 days since being marked as stale.

@github-actions github-actions bot closed this as completed Apr 2, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
bug-unconfirmed stale
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@joaomdmoura @tartakynov @QueryType