Hugging Face's logo

Spaces:
Dovakiins
/
qwerrwe
Build error

App Files Community
winglian commited on
Commit
4fde300
·
unverified ·
1 Parent(s): 3319780

update outputs path so that we can mount workspace to /workspace/data (#1623)

Browse files

* update outputs path so that we can mount workspace to /workspace/data

* fix ln order

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. docker/Dockerfile-cloud +3 -1
  2. examples/cerebras/btlm-ft.yml +1 -1
  3. examples/cerebras/qlora.yml +1 -1
  4. examples/code-llama/13b/lora.yml +1 -1
  5. examples/code-llama/13b/qlora.yml +1 -1
  6. examples/code-llama/34b/lora.yml +1 -1
  7. examples/code-llama/34b/qlora.yml +1 -1
  8. examples/code-llama/7b/lora.yml +1 -1
  9. examples/code-llama/7b/qlora.yml +1 -1
  10. examples/colab-notebooks/colab-axolotl-example.ipynb +1 -1
  11. examples/dbrx/16bit-lora.yaml +1 -1
  12. examples/dbrx/8bit-lora.yaml +1 -1
  13. examples/dbrx/fft-ds-zero3.yaml +1 -1
  14. examples/falcon/config-7b-lora.yml +1 -1
  15. examples/falcon/config-7b-qlora.yml +1 -1
  16. examples/falcon/config-7b.yml +1 -1
  17. examples/gemma/qlora.yml +1 -1
  18. examples/gptj/qlora.yml +1 -1
  19. examples/jamba/qlora.yaml +1 -1
  20. examples/jamba/qlora_deepspeed.yaml +1 -1
  21. examples/jeopardy-bot/config.yml +1 -1
  22. examples/llama-2/fft_optimized.yml +1 -1
  23. examples/llama-2/gptq-lora.yml +1 -1
  24. examples/llama-2/lisa.yml +1 -1
  25. examples/llama-2/loftq.yml +1 -1
  26. examples/llama-2/lora.yml +1 -1
  27. examples/llama-2/qlora-fsdp.yml +1 -1
  28. examples/llama-2/qlora.yml +1 -1
  29. examples/llama-2/relora.yml +1 -1
  30. examples/llama-3/fft-8b.yaml +1 -1
  31. examples/llama-3/lora-8b.yml +1 -1
  32. examples/llama-3/qlora-fsdp-70b.yaml +1 -1
  33. examples/llama-3/qlora.yml +1 -1
  34. examples/mamba/config.yml +1 -1
  35. examples/mistral/bigstral-ds-zero3.yaml +1 -1
  36. examples/mistral/config.yml +1 -1
  37. examples/mistral/lora-mps.yml +1 -1
  38. examples/mistral/lora.yml +1 -1
  39. examples/mistral/mistral-qlora-fsdp.yml +1 -1
  40. examples/mistral/mistral-qlora-orpo.yml +1 -1
  41. examples/mistral/mixtral-8x22b-qlora-fsdp.yml +1 -1
  42. examples/mistral/mixtral-qlora-fsdp.yml +1 -1
  43. examples/mistral/mixtral.yml +1 -1
  44. examples/mistral/mixtral_22.yml +1 -1
  45. examples/mistral/qlora.yml +1 -1
  46. examples/mpt-7b/config.yml +1 -1
  47. examples/openllama-3b/config.yml +1 -1
  48. examples/openllama-3b/lora.yml +1 -1
  49. examples/openllama-3b/qlora.yml +1 -1
  50. examples/phi/phi-ft.yml +1 -1
docker/Dockerfile-cloud CHANGED
@@ -21,7 +21,9 @@ RUN apt install --yes --no-install-recommends openssh-server tmux && \
21
  printf "\n[[ -z \"\$TMUX\" ]] && { tmux attach-session -t ssh_tmux || tmux new-session -s ssh_tmux; exit; }\n" >> ~/.bashrc && \
22
  printf "[ ! -z \"\$TERM\" -a -r /etc/motd ] && cat /etc/motd\n" >> ~/.bashrc && \
23
  chmod +x /workspace/axolotl/scripts/cloud-entrypoint.sh && \
24
- chmod +x /root/cloud-entrypoint.sh
 
 
25
 
26
  ENTRYPOINT ["/root/cloud-entrypoint.sh"]
27
  CMD ["sleep", "infinity"]
 
21
  printf "\n[[ -z \"\$TMUX\" ]] && { tmux attach-session -t ssh_tmux || tmux new-session -s ssh_tmux; exit; }\n" >> ~/.bashrc && \
22
  printf "[ ! -z \"\$TERM\" -a -r /etc/motd ] && cat /etc/motd\n" >> ~/.bashrc && \
23
  chmod +x /workspace/axolotl/scripts/cloud-entrypoint.sh && \
24
+ chmod +x /root/cloud-entrypoint.sh && \
25
+ mkdir -p /workspace/data/axolotl-artifacts && \
26
+ ln -sf /workspace/data/axolotl-artifacts /workspace/axolotl/outputs
27
 
28
  ENTRYPOINT ["/root/cloud-entrypoint.sh"]
29
  CMD ["sleep", "infinity"]
examples/cerebras/btlm-ft.yml CHANGED
@@ -38,7 +38,7 @@ wandb_watch:
38
  wandb_name:
39
  wandb_log_model:
40
 
41
- output_dir: btlm-out
42
  gradient_accumulation_steps: 1
43
  micro_batch_size: 1
44
  num_epochs: 1
 
38
  wandb_name:
39
  wandb_log_model:
40
 
41
+ output_dir: ./outputs/btlm-out
42
  gradient_accumulation_steps: 1
43
  micro_batch_size: 1
44
  num_epochs: 1
examples/cerebras/qlora.yml CHANGED
@@ -25,7 +25,7 @@ wandb_entity:
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
- output_dir: ./qlora-out
29
  batch_size: 4
30
  micro_batch_size: 4
31
  num_epochs: 2
 
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
+ output_dir: ./outputs/qlora-out
29
  batch_size: 4
30
  micro_batch_size: 4
31
  num_epochs: 2
examples/code-llama/13b/lora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/code-llama/13b/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/code-llama/34b/lora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/code-llama/34b/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/code-llama/7b/lora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/code-llama/7b/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/colab-notebooks/colab-axolotl-example.ipynb CHANGED
@@ -84,7 +84,7 @@
84
  " type: alpaca\n",
85
  "dataset_prepared_path:\n",
86
  "val_set_size: 0.05\n",
87
- "output_dir: ./qlora-out\n",
88
  "\n",
89
  "adapter: qlora\n",
90
  "lora_model_dir:\n",
 
84
  " type: alpaca\n",
85
  "dataset_prepared_path:\n",
86
  "val_set_size: 0.05\n",
87
+ "output_dir: ./outputs/qlora-out\n",
88
  "\n",
89
  "adapter: qlora\n",
90
  "lora_model_dir:\n",
examples/dbrx/16bit-lora.yaml CHANGED
@@ -10,7 +10,7 @@ datasets:
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
- output_dir: ./out
14
 
15
  sequence_len: 512
16
  sample_packing: false
 
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
+ output_dir: ./outputs/out
14
 
15
  sequence_len: 512
16
  sample_packing: false
examples/dbrx/8bit-lora.yaml CHANGED
@@ -10,7 +10,7 @@ datasets:
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
- output_dir: ./out
14
 
15
  sequence_len: 512
16
  sample_packing: false
 
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
+ output_dir: ./outputs/out
14
 
15
  sequence_len: 512
16
  sample_packing: false
examples/dbrx/fft-ds-zero3.yaml CHANGED
@@ -10,7 +10,7 @@ datasets:
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
- output_dir: ./out
14
 
15
  sequence_len: 512
16
  sample_packing: false
 
10
  type: alpaca
11
  dataset_prepared_path: last_run_prepared
12
  val_set_size: 0.0
13
+ output_dir: ./outputs/out
14
 
15
  sequence_len: 512
16
  sample_packing: false
examples/falcon/config-7b-lora.yml CHANGED
@@ -28,7 +28,7 @@ wandb_entity:
28
  wandb_watch:
29
  wandb_name:
30
  wandb_log_model:
31
- output_dir: ./falcon-7b
32
  batch_size: 2
33
  micro_batch_size: 1
34
  num_epochs: 4
 
28
  wandb_watch:
29
  wandb_name:
30
  wandb_log_model:
31
+ output_dir: ./outputs/falcon-7b
32
  batch_size: 2
33
  micro_batch_size: 1
34
  num_epochs: 4
examples/falcon/config-7b-qlora.yml CHANGED
@@ -42,7 +42,7 @@ wandb_entity:
42
  wandb_watch:
43
  wandb_name:
44
  wandb_log_model:
45
- output_dir: ./qlora-out
46
 
47
  # QLoRA paper Table 9
48
  # - 16 for 7b & 13b
 
42
  wandb_watch:
43
  wandb_name:
44
  wandb_log_model:
45
+ output_dir: ./outputs/qlora-out
46
 
47
  # QLoRA paper Table 9
48
  # - 16 for 7b & 13b
examples/falcon/config-7b.yml CHANGED
@@ -28,7 +28,7 @@ wandb_entity:
28
  wandb_watch:
29
  wandb_name:
30
  wandb_log_model:
31
- output_dir: ./falcon-7b
32
  batch_size: 2
33
  micro_batch_size: 1
34
  num_epochs: 4
 
28
  wandb_watch:
29
  wandb_name:
30
  wandb_log_model:
31
+ output_dir: ./outputs/falcon-7b
32
  batch_size: 2
33
  micro_batch_size: 1
34
  num_epochs: 4
examples/gemma/qlora.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  - path: mhenrichsen/alpaca_2k_test
13
  type: alpaca
14
  val_set_size: 0.1
15
- output_dir: ./out
16
 
17
  adapter: qlora
18
  lora_r: 32
 
12
  - path: mhenrichsen/alpaca_2k_test
13
  type: alpaca
14
  val_set_size: 0.1
15
+ output_dir: ./outputs/out
16
 
17
  adapter: qlora
18
  lora_r: 32
examples/gptj/qlora.yml CHANGED
@@ -23,7 +23,7 @@ wandb_entity:
23
  wandb_watch:
24
  wandb_name:
25
  wandb_log_model:
26
- output_dir: ./qlora-out
27
  gradient_accumulation_steps: 2
28
  micro_batch_size: 2
29
  num_epochs: 2
 
23
  wandb_watch:
24
  wandb_name:
25
  wandb_log_model:
26
+ output_dir: ./outputs/qlora-out
27
  gradient_accumulation_steps: 2
28
  micro_batch_size: 2
29
  num_epochs: 2
examples/jamba/qlora.yaml CHANGED
@@ -10,7 +10,7 @@ datasets:
10
  type: alpaca
11
  dataset_prepared_path:
12
  val_set_size: 0.0
13
- output_dir: ./out
14
 
15
  sequence_len: 4096
16
  sample_packing: false
 
10
  type: alpaca
11
  dataset_prepared_path:
12
  val_set_size: 0.0
13
+ output_dir: ./outputs/out
14
 
15
  sequence_len: 4096
16
  sample_packing: false
examples/jamba/qlora_deepspeed.yaml CHANGED
@@ -10,7 +10,7 @@ datasets:
10
  type: alpaca
11
  dataset_prepared_path:
12
  val_set_size: 0.0
13
- output_dir: ./out
14
 
15
  sequence_len: 4096
16
  sample_packing: false
 
10
  type: alpaca
11
  dataset_prepared_path:
12
  val_set_size: 0.0
13
+ output_dir: ./outputs/out
14
 
15
  sequence_len: 4096
16
  sample_packing: false
examples/jeopardy-bot/config.yml CHANGED
@@ -21,7 +21,7 @@ wandb_entity:
21
  wandb_watch:
22
  wandb_name:
23
  wandb_log_model:
24
- output_dir: ./jeopardy-bot-7b
25
  gradient_accumulation_steps: 1
26
  micro_batch_size: 1
27
  num_epochs: 4
 
21
  wandb_watch:
22
  wandb_name:
23
  wandb_log_model:
24
+ output_dir: ./outputs/jeopardy-bot-7b
25
  gradient_accumulation_steps: 1
26
  micro_batch_size: 1
27
  num_epochs: 4
examples/llama-2/fft_optimized.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
- output_dir: ./out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/llama-2/gptq-lora.yml CHANGED
@@ -33,7 +33,7 @@ wandb_project:
33
  wandb_watch:
34
  wandb_name:
35
  wandb_log_model:
36
- output_dir: ./model-out
37
  gradient_accumulation_steps: 1
38
  micro_batch_size: 1
39
  num_epochs: 4
 
33
  wandb_watch:
34
  wandb_name:
35
  wandb_log_model:
36
+ output_dir: ./outputs/model-out
37
  gradient_accumulation_steps: 1
38
  micro_batch_size: 1
39
  num_epochs: 4
examples/llama-2/lisa.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
- output_dir: ./lisa-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lisa-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/llama-2/loftq.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/llama-2/lora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/llama-2/qlora-fsdp.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/llama-2/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/llama-2/relora.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  type: alpaca
13
  dataset_prepared_path:
14
  val_set_size: 0.05
15
- output_dir: ./relora-out
16
 
17
  adapter: qlora
18
  lora_model_dir:
 
12
  type: alpaca
13
  dataset_prepared_path:
14
  val_set_size: 0.05
15
+ output_dir: ./outputs/relora-out
16
 
17
  adapter: qlora
18
  lora_model_dir:
examples/llama-3/fft-8b.yaml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
- output_dir: ./out
15
 
16
  sequence_len: 8192
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/out
15
 
16
  sequence_len: 8192
17
  sample_packing: true
examples/llama-3/lora-8b.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/lora-out
15
 
16
  sequence_len: 4096
17
  sample_packing: true
examples/llama-3/qlora-fsdp-70b.yaml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
- output_dir: ./out/qlora-llama3-70b
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/out/qlora-llama3-70b
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/llama-3/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/mamba/config.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  type: alpaca
13
  dataset_prepared_path:
14
  val_set_size: 0.0
15
- output_dir: ./out
16
 
17
  sequence_len: 2048
18
  sample_packing: false
 
12
  type: alpaca
13
  dataset_prepared_path:
14
  val_set_size: 0.0
15
+ output_dir: ./outputs/out
16
 
17
  sequence_len: 2048
18
  sample_packing: false
examples/mistral/bigstral-ds-zero3.yaml CHANGED
@@ -23,7 +23,7 @@ datasets:
23
  type: alpaca
24
  dataset_prepared_path: last_run_prepared
25
  val_set_size: 0.05
26
- output_dir: ./out
27
 
28
  sequence_len: 2048
29
  sample_packing: true
 
23
  type: alpaca
24
  dataset_prepared_path: last_run_prepared
25
  val_set_size: 0.05
26
+ output_dir: ./outputs/out
27
 
28
  sequence_len: 2048
29
  sample_packing: true
examples/mistral/config.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
- output_dir: ./out
15
 
16
  sequence_len: 8192
17
  sample_packing: true
 
11
  type: alpaca
12
  dataset_prepared_path:
13
  val_set_size: 0.05
14
+ output_dir: ./outputs/out
15
 
16
  sequence_len: 8192
17
  sample_packing: true
examples/mistral/lora-mps.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0
14
- output_dir: ./lora-out
15
  eval_sample_packing: false
16
 
17
  adapter: lora
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0
14
+ output_dir: ./outputs/lora-out
15
  eval_sample_packing: false
16
 
17
  adapter: lora
examples/mistral/lora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.1
14
- output_dir: ./lora-out
15
 
16
  adapter: lora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.1
14
+ output_dir: ./outputs/lora-out
15
 
16
  adapter: lora
17
  lora_model_dir:
examples/mistral/mistral-qlora-fsdp.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.02
15
- output_dir: ./qlora-out
16
 
17
  model_config:
18
  output_router_logits: true
 
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.02
15
+ output_dir: ./outputs/qlora-out
16
 
17
  model_config:
18
  output_router_logits: true
examples/mistral/mistral-qlora-orpo.yml CHANGED
@@ -16,7 +16,7 @@ datasets:
16
  type: chat_template.argilla
17
  dataset_prepared_path: last_run_prepared
18
  val_set_size: 0.1
19
- output_dir: ./mistral-qlora-orpo-out
20
 
21
  adapter: qlora
22
  lora_model_dir:
 
16
  type: chat_template.argilla
17
  dataset_prepared_path: last_run_prepared
18
  val_set_size: 0.1
19
+ output_dir: ./outputs/mistral-qlora-orpo-out
20
 
21
  adapter: qlora
22
  lora_model_dir:
examples/mistral/mixtral-8x22b-qlora-fsdp.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.02
14
- output_dir: ./qlora-out
15
 
16
  model_config:
17
  output_router_logits: true
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.02
14
+ output_dir: ./outputs/qlora-out
15
 
16
  model_config:
17
  output_router_logits: true
examples/mistral/mixtral-qlora-fsdp.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.02
15
- output_dir: ./qlora-out
16
 
17
  model_config:
18
  output_router_logits: true
 
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.02
15
+ output_dir: ./outputs/qlora-out
16
 
17
  model_config:
18
  output_router_logits: true
examples/mistral/mixtral.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.0
15
- output_dir: ./qlora-out
16
 
17
  ## You can optionally freeze the entire model and unfreeze a subset of parameters
18
  unfrozen_parameters:
 
12
  type: alpaca
13
  dataset_prepared_path: last_run_prepared
14
  val_set_size: 0.0
15
+ output_dir: ./outputs/qlora-out
16
 
17
  ## You can optionally freeze the entire model and unfreeze a subset of parameters
18
  unfrozen_parameters:
examples/mistral/mixtral_22.yml CHANGED
@@ -21,7 +21,7 @@ model_config:
21
  datasets:
22
  - path: yahma/alpaca-cleaned
23
  type: alpaca
24
- output_dir: ./out
25
 
26
  sequence_len: 8000
27
  sample_packing: true
 
21
  datasets:
22
  - path: yahma/alpaca-cleaned
23
  type: alpaca
24
+ output_dir: ./outputs/out
25
 
26
  sequence_len: 8000
27
  sample_packing: true
examples/mistral/qlora.yml CHANGED
@@ -11,7 +11,7 @@ datasets:
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.1
14
- output_dir: ./qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
 
11
  type: alpaca
12
  dataset_prepared_path: last_run_prepared
13
  val_set_size: 0.1
14
+ output_dir: ./outputs/qlora-out
15
 
16
  adapter: qlora
17
  lora_model_dir:
examples/mpt-7b/config.yml CHANGED
@@ -23,7 +23,7 @@ wandb_entity:
23
  wandb_watch:
24
  wandb_name:
25
  wandb_log_model:
26
- output_dir: ./mpt-alpaca-7b
27
  gradient_accumulation_steps: 1
28
  micro_batch_size: 1
29
  num_epochs: 4
 
23
  wandb_watch:
24
  wandb_name:
25
  wandb_log_model:
26
+ output_dir: ./outputs/mpt-alpaca-7b
27
  gradient_accumulation_steps: 1
28
  micro_batch_size: 1
29
  num_epochs: 4
examples/openllama-3b/config.yml CHANGED
@@ -25,7 +25,7 @@ wandb_entity:
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
- output_dir: ./openllama-out
29
  gradient_accumulation_steps: 1
30
  micro_batch_size: 1
31
  num_epochs: 4
 
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
+ output_dir: ./outputs/openllama-out
29
  gradient_accumulation_steps: 1
30
  micro_batch_size: 1
31
  num_epochs: 4
examples/openllama-3b/lora.yml CHANGED
@@ -31,7 +31,7 @@ wandb_entity:
31
  wandb_watch:
32
  wandb_name:
33
  wandb_log_model:
34
- output_dir: ./lora-out
35
  gradient_accumulation_steps: 1
36
  micro_batch_size: 2
37
  num_epochs: 4
 
31
  wandb_watch:
32
  wandb_name:
33
  wandb_log_model:
34
+ output_dir: ./outputs/lora-out
35
  gradient_accumulation_steps: 1
36
  micro_batch_size: 2
37
  num_epochs: 4
examples/openllama-3b/qlora.yml CHANGED
@@ -25,7 +25,7 @@ wandb_entity:
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
- output_dir: ./qlora-out
29
  gradient_accumulation_steps: 1
30
  micro_batch_size: 2
31
  num_epochs: 4
 
25
  wandb_watch:
26
  wandb_name:
27
  wandb_log_model:
28
+ output_dir: ./outputs/qlora-out
29
  gradient_accumulation_steps: 1
30
  micro_batch_size: 2
31
  num_epochs: 4
examples/phi/phi-ft.yml CHANGED
@@ -12,7 +12,7 @@ datasets:
12
 
13
  dataset_prepared_path:
14
  val_set_size: 0.05
15
- output_dir: ./phi-sft-out
16
 
17
  sequence_len: 2048
18
  sample_packing: true
 
12
 
13
  dataset_prepared_path:
14
  val_set_size: 0.05
15
+ output_dir: ./outputs/phi-sft-out
16
 
17
  sequence_len: 2048
18
  sample_packing: true