migrate example configs to canonical attn_implementation

2026-04-23 22:15:07 +00:00
parent 2d64d009d8
commit 39226623d2
222 changed files with 209 additions and 243 deletions
--- a/examples/LiquidAI/lfm2-350m-fft.yaml
+++ b/examples/LiquidAI/lfm2-350m-fft.yaml
@@ -39,7 +39,7 @@ tf32: true
 gradient_checkpointing: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 2
--- a/examples/LiquidAI/lfm2-8b-a1b-lora.yaml
+++ b/examples/LiquidAI/lfm2-8b-a1b-lora.yaml
@@ -48,7 +48,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 2
--- a/examples/LiquidAI/lfm2-vl-lora.yaml
+++ b/examples/LiquidAI/lfm2-vl-lora.yaml
@@ -50,8 +50,7 @@ tf32: true

 gradient_checkpointing: true
 logging_steps: 1
-flash_attention: true
-eager_attention:
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/alst/llama3-8b-deepspeed-alst.yaml
+++ b/examples/alst/llama3-8b-deepspeed-alst.yaml
@@ -39,7 +39,7 @@ activation_offloading: legacy

 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_steps: 100
 saves_per_epoch: 1
--- a/examples/alst/llama3-8b-fsdp2-alst.yaml
+++ b/examples/alst/llama3-8b-fsdp2-alst.yaml
@@ -39,7 +39,7 @@ activation_offloading: legacy

 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_steps: 100
 saves_per_epoch: 1
--- a/examples/apertus/apertus-8b-qlora.yaml
+++ b/examples/apertus/apertus-8b-qlora.yaml
@@ -55,7 +55,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/arcee/afm-4.5b-qlora.yaml
+++ b/examples/arcee/afm-4.5b-qlora.yaml
@@ -55,7 +55,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/archived/cerebras/btlm-ft.yml
+++ b/examples/archived/cerebras/btlm-ft.yml
@@ -59,8 +59,7 @@ gradient_checkpointing: false
 resume_from_checkpoint:
 logging_steps: 1

-flash_attention: true
-sdp_attention:
+attn_implementation: flash_attention_2
 flash_optimum:

 gptq_groupsize:
--- a/examples/archived/cerebras/qlora.yml
+++ b/examples/archived/cerebras/qlora.yml
@@ -39,8 +39,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/code-llama/13b/lora.yml
+++ b/examples/archived/code-llama/13b/lora.yml
@@ -45,7 +45,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/code-llama/13b/qlora.yml
+++ b/examples/archived/code-llama/13b/qlora.yml
@@ -46,7 +46,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/code-llama/34b/lora.yml
+++ b/examples/archived/code-llama/34b/lora.yml
@@ -45,7 +45,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/code-llama/34b/qlora.yml
+++ b/examples/archived/code-llama/34b/qlora.yml
@@ -46,7 +46,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/code-llama/7b/lora.yml
+++ b/examples/archived/code-llama/7b/lora.yml
@@ -45,7 +45,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/code-llama/7b/qlora.yml
+++ b/examples/archived/code-llama/7b/qlora.yml
@@ -46,7 +46,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/dbrx/16bit-lora.yaml
+++ b/examples/archived/dbrx/16bit-lora.yaml
@@ -52,7 +52,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/archived/dbrx/8bit-lora.yaml
+++ b/examples/archived/dbrx/8bit-lora.yaml
@@ -55,7 +55,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/archived/dbrx/fft-ds-zero3.yaml
+++ b/examples/archived/dbrx/fft-ds-zero3.yaml
@@ -39,7 +39,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/archived/deepcoder/deepcoder-14B-preview-lora.yml
+++ b/examples/archived/deepcoder/deepcoder-14B-preview-lora.yml
@@ -45,7 +45,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/archived/falcon/config-7b-lora.yml
+++ b/examples/archived/falcon/config-7b-lora.yml
@@ -43,8 +43,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/falcon/config-7b-qlora.yml
+++ b/examples/archived/falcon/config-7b-qlora.yml
@@ -73,8 +73,7 @@ early_stopping_patience: 3
 resume_from_checkpoint:
 auto_resume_from_checkpoints: true
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/falcon/config-7b.yml
+++ b/examples/archived/falcon/config-7b.yml
@@ -40,8 +40,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/gemma/qlora.yml
+++ b/examples/archived/gemma/qlora.yml
@@ -47,7 +47,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/gptj/qlora.yml
+++ b/examples/archived/gptj/qlora.yml
@@ -36,8 +36,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/jeopardy-bot/config.yml
+++ b/examples/archived/jeopardy-bot/config.yml
@@ -37,8 +37,7 @@ bf16: auto
 tf32: true
 resume_from_checkpoint:
 logging_steps: 5
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/mpt-7b/config.yml
+++ b/examples/archived/mpt-7b/config.yml
@@ -39,7 +39,6 @@ bf16: auto
 tf32: true
 resume_from_checkpoint:
 logging_steps: 5
-flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/openllama-3b/config.yml
+++ b/examples/archived/openllama-3b/config.yml
@@ -39,7 +39,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/openllama-3b/lora.yml
+++ b/examples/archived/openllama-3b/lora.yml
@@ -47,7 +47,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/openllama-3b/qlora.yml
+++ b/examples/archived/openllama-3b/qlora.yml
@@ -40,7 +40,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/qwen/lora.yml
+++ b/examples/archived/qwen/lora.yml
@@ -47,7 +47,6 @@ tf32: false
 gradient_checkpointing: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention:

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/qwen/qlora.yml
+++ b/examples/archived/qwen/qlora.yml
@@ -47,7 +47,6 @@ tf32: false
 gradient_checkpointing: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention:

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/qwen/qwen2-moe-lora.yaml
+++ b/examples/archived/qwen/qwen2-moe-lora.yaml
@@ -43,7 +43,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/qwen/qwen2-moe-qlora.yaml
+++ b/examples/archived/qwen/qwen2-moe-qlora.yaml
@@ -46,7 +46,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/redpajama/config-3b.yml
+++ b/examples/archived/redpajama/config-3b.yml
@@ -40,7 +40,6 @@ bf16: auto
 tf32: true
 resume_from_checkpoint:
 logging_steps: 5
-flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/replit-3b/config-lora.yml
+++ b/examples/archived/replit-3b/config-lora.yml
@@ -38,7 +38,6 @@ tf32: true
 gradient_checkpointing:
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/stablelm-2/1.6b/fft.yml
+++ b/examples/archived/stablelm-2/1.6b/fft.yml
@@ -44,7 +44,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 flash_attn_cross_entropy: false
 flash_attn_rms_norm: true
 flash_attn_fuse_mlp: true
--- a/examples/archived/stablelm-2/1.6b/lora.yml
+++ b/examples/archived/stablelm-2/1.6b/lora.yml
@@ -47,7 +47,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 flash_attn_cross_entropy: false
 flash_attn_rms_norm: true

--- a/examples/archived/starcoder2/qlora.yml
+++ b/examples/archived/starcoder2/qlora.yml
@@ -46,7 +46,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/tiny-llama/lora-mps.yml
+++ b/examples/archived/tiny-llama/lora-mps.yml
@@ -47,7 +47,6 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: false

 warmup_ratio: 0.1
 evals_per_epoch: 0
--- a/examples/archived/tiny-llama/lora.yml
+++ b/examples/archived/tiny-llama/lora.yml
@@ -45,7 +45,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/tiny-llama/pretrain.yml
+++ b/examples/archived/tiny-llama/pretrain.yml
@@ -36,7 +36,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/archived/tiny-llama/qlora.yml
+++ b/examples/archived/tiny-llama/qlora.yml
@@ -47,7 +47,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/archived/xgen-7b/xgen-7b-8k-qlora.yml
+++ b/examples/archived/xgen-7b/xgen-7b-8k-qlora.yml
@@ -71,8 +71,7 @@ early_stopping_patience: 3
 resume_from_checkpoint:
 auto_resume_from_checkpoints: true
 logging_steps: 1
-xformers_attention: true
-flash_attention:
+attn_implementation: xformers
 gptq_groupsize:
 gptq_model_v1:
 warmup_ratio: 0.1
--- a/examples/archived/yi-34B-chat/qlora.yml
+++ b/examples/archived/yi-34B-chat/qlora.yml
@@ -10,7 +10,7 @@ load_in_4bit: true
 sequence_len: 1024
 bf16: auto
 tf32: false
-flash_attention: true
+attn_implementation: flash_attention_2
 special_tokens:
  bos_token: "<|startoftext|>"
  eos_token: "<|endoftext|>"
--- a/examples/cohere/command-r-7b-qlora.yml
+++ b/examples/cohere/command-r-7b-qlora.yml
@@ -48,7 +48,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/deepcogito/cogito-v1-preview-llama-3B-lora.yml
+++ b/examples/deepcogito/cogito-v1-preview-llama-3B-lora.yml
@@ -45,7 +45,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/deepcogito/cogito-v1-preview-qwen-14B-lora.yml
+++ b/examples/deepcogito/cogito-v1-preview-qwen-14B-lora.yml
@@ -45,7 +45,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/deepseek-v2/fft-fsdp-16b.yaml
+++ b/examples/deepseek-v2/fft-fsdp-16b.yaml
@@ -35,7 +35,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 2
--- a/examples/deepseek-v2/qlora-fsdp-2_5.yaml
+++ b/examples/deepseek-v2/qlora-fsdp-2_5.yaml
@@ -59,7 +59,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 2
--- a/examples/devstral/devstral-small-qlora.yml
+++ b/examples/devstral/devstral-small-qlora.yml
@@ -51,7 +51,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 scaling_softmax: true

 loss_watchdog_threshold: 5.0
--- a/examples/distributed-parallel/llama-3_1-8b-hsdp-tp.yaml
+++ b/examples/distributed-parallel/llama-3_1-8b-hsdp-tp.yaml
@@ -29,7 +29,7 @@ output_dir: ./outputs/ndp-out/

 sequence_len: 2048
 sample_packing: true
-flash_attention: true
+attn_implementation: flash_attention_2

 gradient_accumulation_steps: 1
 micro_batch_size: 1
--- a/examples/distributed-parallel/qwen3-8b-fsdp-tp-cp.yaml
+++ b/examples/distributed-parallel/qwen3-8b-fsdp-tp-cp.yaml
@@ -26,7 +26,7 @@ output_dir: ./outputs/ndp-out/

 sequence_len: 8192
 sample_packing: true
-flash_attention: true
+attn_implementation: flash_attention_2

 gradient_accumulation_steps: 1
 micro_batch_size: 1  # must be 1 when using context parallel
--- a/examples/eaft/eaft-example.yml
+++ b/examples/eaft/eaft-example.yml
@@ -65,8 +65,7 @@ early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
 logging_steps: 1
-xformers_attention:
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 weight_decay: 0.0
--- a/examples/ebft/llama-1b-ebft-opencode-novllm.yaml
+++ b/examples/ebft/llama-1b-ebft-opencode-novllm.yaml
@@ -46,7 +46,7 @@ lora_dropout: 0.05
 lora_target_linear: true

 bf16: auto
-flash_attention: true
+attn_implementation: flash_attention_2
 gradient_checkpointing: true

 special_tokens:
--- a/examples/ebft/llama-1b-ebft-opencode.yaml
+++ b/examples/ebft/llama-1b-ebft-opencode.yaml
@@ -66,7 +66,7 @@ lora_target_linear: true

 # --- Hardware ---
 bf16: auto
-flash_attention: true
+attn_implementation: flash_attention_2
 gradient_checkpointing: true

 special_tokens:
--- a/examples/ebft/llama-1b-ebft-strided-structured.yaml
+++ b/examples/ebft/llama-1b-ebft-strided-structured.yaml
@@ -47,8 +47,7 @@ lora_dropout: 0.05
 lora_target_linear: true

 bf16: auto
-flash_attention: false  # strided EBFT overrides to flex_attention (or eager fallback) at runtime
-flex_attention: true    # fused flex_attention kernel compiles itself; don't set torch_compile: true
+attn_implementation: flex_attention
                        # (full-model compile conflicts with gradient checkpointing + flex_attention)
 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
--- a/examples/ebft/llama-1b-ebft-strided.yaml
+++ b/examples/ebft/llama-1b-ebft-strided.yaml
@@ -46,7 +46,6 @@ lora_dropout: 0.05
 lora_target_linear: true

 bf16: auto
-flash_attention: false  # strided EBFT overrides to flex_attention (or eager fallback) at runtime
 gradient_checkpointing: true

 special_tokens:
--- a/examples/ebft/llama-3b-ebft-strided-fft.yaml
+++ b/examples/ebft/llama-3b-ebft-strided-fft.yaml
@@ -48,7 +48,6 @@ lora_target_linear: true

 bf16: auto
 torch_dtype: bfloat16
-flash_attention: false
 gradient_checkpointing: true
 torch_compile: true
 gradient_checkpointing_kwargs:
--- a/examples/ebft/llama-8b-ebft-strided-fft.yaml
+++ b/examples/ebft/llama-8b-ebft-strided-fft.yaml
@@ -41,7 +41,6 @@ warmup_steps: 10
 weight_decay: 0.01

 bf16: auto
-flash_attention: false  # strided EBFT uses flex_attention at runtime
 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
--- a/examples/ebft/qwen35-4b-ebft-structured-async.yaml
+++ b/examples/ebft/qwen35-4b-ebft-structured-async.yaml
@@ -72,7 +72,7 @@ lora_dropout: 0.0
 lora_target_modules: ".*\\.layers\\.(3|7|11|15|19|23|27|31)\\.self_attn\\.(q|k|v|o)_proj|.*\\.mlp\\.(gate|up|down)_proj"

 bf16: auto
-flash_attention: true
+attn_implementation: flash_attention_2
 gradient_checkpointing: true

 special_tokens:
--- a/examples/ebft/qwen35-4b-ebft-structured.yaml
+++ b/examples/ebft/qwen35-4b-ebft-structured.yaml
@@ -63,7 +63,7 @@ lora_dropout: 0.0
 lora_target_modules: ".*\\.layers\\.(3|7|11|15|19|23|27|31)\\.self_attn\\.(q|k|v|o)_proj|.*\\.mlp\\.(gate|up|down)_proj"

 bf16: auto
-flash_attention: true
+attn_implementation: flash_attention_2
 gradient_checkpointing: true

 special_tokens:
--- a/examples/ebft/qwen35-9b-ebft-structured.yaml
+++ b/examples/ebft/qwen35-9b-ebft-structured.yaml
@@ -68,7 +68,7 @@ lora_dropout: 0.0
 lora_target_modules: ".*\\.layers\\.(3|7|11|15|19|23|27|31)\\.self_attn\\.(q|k|v|o)_proj|.*\\.mlp\\.(gate|up|down)_proj"

 bf16: auto
-flash_attention: true
+attn_implementation: flash_attention_2
 gradient_checkpointing: true

 special_tokens:
--- a/examples/falcon-h1/falcon-h1-1b-deep-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-1b-deep-qlora.yaml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/falcon-h1/falcon-h1-1b-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-1b-qlora.yaml
@@ -61,7 +61,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/falcon-h1/falcon-h1-34b-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-34b-qlora.yaml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/falcon-h1/falcon-h1-3b-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-3b-qlora.yaml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/falcon-h1/falcon-h1-500m-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-500m-qlora.yaml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/falcon-h1/falcon-h1-7b-qlora.yaml
+++ b/examples/falcon-h1/falcon-h1-7b-qlora.yaml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/gemma2/qlora.yml
+++ b/examples/gemma2/qlora.yml
@@ -53,7 +53,7 @@ tf32: true
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/gemma2/reward-model.yaml
+++ b/examples/gemma2/reward-model.yaml
@@ -43,7 +43,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/gemma3/gemma-3-1b-qlora.yml
+++ b/examples/gemma3/gemma-3-1b-qlora.yml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/gemma3/gemma-3-270m-qlora.yml
+++ b/examples/gemma3/gemma-3-270m-qlora.yml
@@ -62,7 +62,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/gemma3/gemma-3-4b-qlora.yml
+++ b/examples/gemma3/gemma-3-4b-qlora.yml
@@ -58,8 +58,7 @@ gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
 logging_steps: 1
-flash_attention: true
-eager_attention:
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/gemma3/gemma-3-4b-vision-qlora.yml
+++ b/examples/gemma3/gemma-3-4b-vision-qlora.yml
@@ -55,8 +55,7 @@ gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
 logging_steps: 1
-flash_attention: true
-eager_attention:
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/gemma4/26b-a4b-moe-qlora.yaml
+++ b/examples/gemma4/26b-a4b-moe-qlora.yaml
@@ -84,7 +84,7 @@ activation_offloading: true
 logging_steps: 1

 # FA2 not supported
-sdp_attention: true
+attn_implementation: sdpa

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/gemma4/31b-qlora-flex.yaml
+++ b/examples/gemma4/31b-qlora-flex.yaml
@@ -62,7 +62,7 @@ activation_offloading: true
 logging_steps: 1

 # FA not supported
-flex_attention: true
+attn_implementation: flex_attention

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/gemma4/31b-qlora.yaml
+++ b/examples/gemma4/31b-qlora.yaml
@@ -60,7 +60,7 @@ activation_offloading: true
 logging_steps: 1

 # FA not supported
-sdp_attention: true
+attn_implementation: sdpa

 warmup_ratio: 0.1
 evals_per_epoch: 4
--- a/examples/gemma4/e2b-vision-lora.yaml
+++ b/examples/gemma4/e2b-vision-lora.yaml
@@ -50,7 +50,7 @@ gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
 logging_steps: 1
-sdp_attention: true
+attn_implementation: sdpa

 warmup_ratio: 0.1
 weight_decay: 0.0
--- a/examples/glm4/qlora-32b.yaml
+++ b/examples/glm4/qlora-32b.yaml
@@ -50,7 +50,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 loss_watchdog_threshold: 5.0
 loss_watchdog_patience: 3
--- a/examples/glm45/glm-45-air-qlora.yaml
+++ b/examples/glm45/glm-45-air-qlora.yaml
@@ -55,7 +55,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/glm46v/glm-4-6v-flash-ddp.yaml
+++ b/examples/glm46v/glm-4-6v-flash-ddp.yaml
@@ -45,7 +45,7 @@ gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
 logging_steps: 1
-sdp_attention: true
+attn_implementation: sdpa

 warmup_ratio: 0.1
 evals_per_epoch: 0
--- a/examples/glm46v/glm-4-6v-flash-qlora.yaml
+++ b/examples/glm46v/glm-4-6v-flash-qlora.yaml
@@ -42,7 +42,7 @@ tf32: false

 gradient_checkpointing: true
 logging_steps: 1
-sdp_attention: true
+attn_implementation: sdpa

 warmup_ratio: 0.1
 evals_per_epoch: 0
--- a/examples/glm47-flash/lora.yaml
+++ b/examples/glm47-flash/lora.yaml
@@ -58,7 +58,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/glm47-flash/lora_fsdp.yaml
+++ b/examples/glm47-flash/lora_fsdp.yaml
@@ -57,7 +57,7 @@ tf32: false

 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/glm47-flash/qlora.yaml
+++ b/examples/glm47-flash/qlora.yaml
@@ -58,7 +58,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/glm47-flash/qlora_fsdp.yaml
+++ b/examples/glm47-flash/qlora_fsdp.yaml
@@ -57,7 +57,7 @@ tf32: false

 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/gpt-oss/gpt-oss-120b-fft-fsdp2-offload.yaml
+++ b/examples/gpt-oss/gpt-oss-120b-fft-fsdp2-offload.yaml
@@ -47,7 +47,7 @@ learning_rate: 2e-5
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/gpt-oss/gpt-oss-20b-fft-deepspeed-zero3.yaml
+++ b/examples/gpt-oss/gpt-oss-20b-fft-deepspeed-zero3.yaml
@@ -43,7 +43,7 @@ learning_rate: 2e-5
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/gpt-oss/gpt-oss-20b-fft-fsdp2-offload.yaml
+++ b/examples/gpt-oss/gpt-oss-20b-fft-fsdp2-offload.yaml
@@ -44,7 +44,7 @@ learning_rate: 2e-5
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/gpt-oss/gpt-oss-20b-fft-fsdp2.yaml
+++ b/examples/gpt-oss/gpt-oss-20b-fft-fsdp2.yaml
@@ -43,7 +43,7 @@ learning_rate: 2e-5
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/gpt-oss/gpt-oss-20b-sft-lora-singlegpu.yaml
+++ b/examples/gpt-oss/gpt-oss-20b-sft-lora-singlegpu.yaml
@@ -56,7 +56,7 @@ learning_rate: 2e-4
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/gpt-oss/gpt-oss-safeguard-20b-sft-lora-singlegpu.yaml
+++ b/examples/gpt-oss/gpt-oss-safeguard-20b-sft-lora-singlegpu.yaml
@@ -56,7 +56,7 @@ learning_rate: 2e-4
 bf16: true
 tf32: true

-flash_attention: true
+attn_implementation: flash_attention_2
 attn_implementation: kernels-community/vllm-flash-attn3  # this is not needed if using flash_attn >= 2.8.3

 gradient_checkpointing: true
--- a/examples/granite4/granite-4.0-tiny-fft.yaml
+++ b/examples/granite4/granite-4.0-tiny-fft.yaml
@@ -36,7 +36,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/hunyuan/hunyuan-v1-dense-qlora.yaml
+++ b/examples/hunyuan/hunyuan-v1-dense-qlora.yaml
@@ -55,7 +55,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/internvl3_5/internvl3_5-8b-qlora.yml
+++ b/examples/internvl3_5/internvl3_5-8b-qlora.yml
@@ -50,8 +50,7 @@ tf32: true

 gradient_checkpointing: true
 logging_steps: 1
-flash_attention: true
-eager_attention:
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/jamba/qlora.yaml
+++ b/examples/jamba/qlora.yaml
@@ -47,7 +47,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/jamba/qlora_deepspeed.yaml
+++ b/examples/jamba/qlora_deepspeed.yaml
@@ -46,7 +46,7 @@ gradient_checkpointing_kwargs:
  use_reentrant: false
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch:
--- a/examples/jamba/qlora_fsdp_large.yaml
+++ b/examples/jamba/qlora_fsdp_large.yaml
@@ -44,7 +44,7 @@ gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: true
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 warmup_ratio: 0.1
 evals_per_epoch: 1
--- a/examples/kimi-linear/kimi-48b-lora.yaml
+++ b/examples/kimi-linear/kimi-48b-lora.yaml
@@ -65,7 +65,7 @@ early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2

 loss_watchdog_threshold: 5.0
 loss_watchdog_patience: 3
--- a/examples/llama-2/fft_optimized.yml
+++ b/examples/llama-2/fft_optimized.yml
@@ -42,7 +42,7 @@ tf32: false
 gradient_checkpointing: true
 resume_from_checkpoint:
 logging_steps: 1
-flash_attention: true
+attn_implementation: flash_attention_2
 flash_attn_cross_entropy: false
 flash_attn_rms_norm: true
 flash_attn_fuse_mlp: true
--- a/Show More
+++ b/Show More