project-llm-trainer 0.5.9__py3-none-any.whl → 0.5.10__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of project-llm-trainer might be problematic. Click here for more details.

Files changed (12) hide show
  1. llm_trainer/checkpoint.py +6 -5
  2. {project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/METADATA +1 -1
  3. {project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/RECORD +12 -12
  4. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/calc_intermediate_size +0 -0
  5. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/ddp_train +0 -0
  6. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/ds_train +0 -0
  7. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/plot_loss +0 -0
  8. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/plot_lr +0 -0
  9. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/py_train +0 -0
  10. {project_llm_trainer-0.5.9.data → project_llm_trainer-0.5.10.data}/scripts/smart_train +0 -0
  11. {project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/WHEEL +0 -0
  12. {project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/top_level.txt +0 -0
llm_trainer/checkpoint.py CHANGED
@@ -81,12 +81,13 @@ def load_checkpoint(
81
81
  else:
82
82
  checkpoint_name = os.environ.get('CHECKPOINT_NAME', DEFAULT_CHECKPOINT_NAME)
83
83
 
84
- state_dict = torch.load(checkpoint_name, weights_only=True, map_location=device)
85
- raw_model = model.module if isinstance(model, DDP) else model
86
- raw_model.load_state_dict(state_dict['model_state_dict'])
84
+ if os.path.exists(checkpoint_name):
85
+ state_dict = torch.load(checkpoint_name, weights_only=True, map_location=device)
86
+ raw_model = model.module if isinstance(model, DDP) else model
87
+ raw_model.load_state_dict(state_dict['model_state_dict'])
87
88
 
88
- if optimizer:
89
- optimizer.load_state_dict(state_dict['optim_state_dict'])
89
+ if optimizer:
90
+ optimizer.load_state_dict(state_dict['optim_state_dict'])
90
91
 
91
92
 
92
93
  def load_checkpoint_for_eval(
{project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: project_llm_trainer
3
- Version: 0.5.9
3
+ Version: 0.5.10
4
4
  Summary: LLM and VLM trainer
5
5
  Author: qibin
6
6
  Author-email: qibin0506@gmail.com
{project_llm_trainer-0.5.9.dist-info → project_llm_trainer-0.5.10.dist-info}/RECORD RENAMED
@@ -1,5 +1,5 @@
1
1
  llm_trainer/__init__.py,sha256=HWgtTEVeQSnZmEyYQm2K6eFEG4X2QAoigMlB5Z2tcXE,260
2
- llm_trainer/checkpoint.py,sha256=RoRlIB-Qtvl3MyY3g0FbEBHLpFRLnEMZLpEncXOLToQ,4242
2
+ llm_trainer/checkpoint.py,sha256=Wh5CwceIajTgJ9i_mH3I1R9N2nOLFqVFmlEMkTiGcD4,4306
3
3
  llm_trainer/dataset.py,sha256=4QlOo0SFB5816BUYegQjgobUqTUMQvdmZMM_OEAMSjE,4347
4
4
  llm_trainer/dpo_trainer.py,sha256=1A_4QP2_xqM_YeqdXy-0RaMvEL80gim-pgnPQyHww9U,12052
5
5
  llm_trainer/ds_checkpoint.py,sha256=D092fkS1Up4QmpV9YCpqbSzfX_caCAeX-UiOrhOE1I8,1947
@@ -20,14 +20,14 @@ llm_trainer/tools.py,sha256=yF17lp6oOfLe2XJeKDQ1juZcbv-6vFamJSLwEeArduA,2975
20
20
  llm_trainer/train_configs.py,sha256=guV8xkG5TSGvYwFvsQV_mA8mDHLLVhL5L0xo_WMsMME,7347
21
21
  llm_trainer/trainer.py,sha256=U26dZc22nByfTZUzKeEiqqYVexBzgw0ep7N0Z2zIcWI,26141
22
22
  llm_trainer/utils.py,sha256=LWNhyQ0NDEZ9mZtk2Ryvh6EulvHIaUGIflugSpqmeFI,6791
23
- project_llm_trainer-0.5.9.data/scripts/calc_intermediate_size,sha256=AggpgNHokJiJMbEtVdOnolqr_4bH3i1UYuZNEAzC2Gc,460
24
- project_llm_trainer-0.5.9.data/scripts/ddp_train,sha256=x81AasaN2-9TwARFFF1l7iV1LmfMQ0bLw0i_CGbOwSw,299
25
- project_llm_trainer-0.5.9.data/scripts/ds_train,sha256=qL3qc3TcedBCw98UZUjW07ONcErRawLE1HymW2AmscA,265
26
- project_llm_trainer-0.5.9.data/scripts/plot_loss,sha256=MzFcdJESlVr1srj4Td6-AxPGUKkfB_QEcJwm0Bd-5fU,910
27
- project_llm_trainer-0.5.9.data/scripts/plot_lr,sha256=w_7XR_x3KYYyboeOVAeu_I4fveLFI-C0wBmRrNlmWUI,894
28
- project_llm_trainer-0.5.9.data/scripts/py_train,sha256=tOp9TquORQeU8XN5H7OVIk5O0Ypwi34p_GENxTwgwdk,265
29
- project_llm_trainer-0.5.9.data/scripts/smart_train,sha256=Pmt4Q0to4Hoz82iB9uFPZuz7uahNUbfE7FR1940EBy8,716
30
- project_llm_trainer-0.5.9.dist-info/METADATA,sha256=YfFvnbVfUyNMCByDKCJ1rB4Mj0uxGQ2wquSe4QKaiF4,195
31
- project_llm_trainer-0.5.9.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
32
- project_llm_trainer-0.5.9.dist-info/top_level.txt,sha256=LtRFg28i0QIG7iBCD2t095oSco99LCtkijibS9cMGik,12
33
- project_llm_trainer-0.5.9.dist-info/RECORD,,
23
+ project_llm_trainer-0.5.10.data/scripts/calc_intermediate_size,sha256=AggpgNHokJiJMbEtVdOnolqr_4bH3i1UYuZNEAzC2Gc,460
24
+ project_llm_trainer-0.5.10.data/scripts/ddp_train,sha256=x81AasaN2-9TwARFFF1l7iV1LmfMQ0bLw0i_CGbOwSw,299
25
+ project_llm_trainer-0.5.10.data/scripts/ds_train,sha256=qL3qc3TcedBCw98UZUjW07ONcErRawLE1HymW2AmscA,265
26
+ project_llm_trainer-0.5.10.data/scripts/plot_loss,sha256=MzFcdJESlVr1srj4Td6-AxPGUKkfB_QEcJwm0Bd-5fU,910
27
+ project_llm_trainer-0.5.10.data/scripts/plot_lr,sha256=w_7XR_x3KYYyboeOVAeu_I4fveLFI-C0wBmRrNlmWUI,894
28
+ project_llm_trainer-0.5.10.data/scripts/py_train,sha256=tOp9TquORQeU8XN5H7OVIk5O0Ypwi34p_GENxTwgwdk,265
29
+ project_llm_trainer-0.5.10.data/scripts/smart_train,sha256=Pmt4Q0to4Hoz82iB9uFPZuz7uahNUbfE7FR1940EBy8,716
30
+ project_llm_trainer-0.5.10.dist-info/METADATA,sha256=Q2et2T0N_hfkIRE2ShyjFX2aSayCTzk1_W4nvFKN820,196
31
+ project_llm_trainer-0.5.10.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
32
+ project_llm_trainer-0.5.10.dist-info/top_level.txt,sha256=LtRFg28i0QIG7iBCD2t095oSco99LCtkijibS9cMGik,12
33
+ project_llm_trainer-0.5.10.dist-info/RECORD,,