id stringlengths 12 32 | archetype stringclasses 30
values | split stringclasses 1
value | prompt stringlengths 4.71k 5.31k | gold stringlengths 68 186 | meta stringclasses 30
values |
|---|---|---|---|---|---|
train-parallel_batch-004 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 64}, "work": {"fixed_memory_gb": 64.0, "params_per_rank_billions": 4.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-parallel_batch-011 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 25}, "work": {"fixed_memory_gb": 128.0, "params_per_rank_billions": 8.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-parallel_batch-069 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 64}, "work": {"fixed_memory_gb": 128.0, "params_per_rank_billions": 8.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-two_v100_dataloader-041 | two_v100_dataloader | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["PRECISION_UNSUPPORTED_ON_GPU", "DATALOADER_FORK_WITH_CUDA"], "fix": {"dataloader_multiprocessing_context": "spawn", "dtype": "fp16", "use_grad_scaler": true}} | {"archetype": "two_v100_dataloader", "computations": ["R5", "R9"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-packing_time-012 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 1.0}, "work": {"total_padded_tokens": 491520000, "total_real_tokens": 216000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-vocab-007 | vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "EMBEDDING_VOCAB_MISMATCH", "fix": {"embedding_vocab_size": 64032}} | {"archetype": "vocab", "computations": ["R11"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-packing_time-010 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 0.5}, "work": {"total_padded_tokens": 245760000, "total_real_tokens": 72000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-accum_steps_lr-023 | accum_steps_lr | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_STEPS_AND_LR", "fix": {"current_learning_rate": 9.75e-05, "grad_accum_steps": 2, "planned_total_steps": 800}, "work": {"audit_step": 600, "steps_per_epoch": 400}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-shard_size-019 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 12500, "shard_size_gb": 25.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-two_lora_vocab-058 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 32064}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-two_lora_vocab-009 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-accum_and_steps-049 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 8, "planned_total_steps": 300}, "work": {"steps_per_epoch": 100}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-shard_size-014 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 25000, "shard_size_gb": 40.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-lr_at_step-021 | lr_at_step | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "LR_VALUE_AT_STEP", "fix": {"current_learning_rate": 0.0002325}, "work": {"step_fraction": 0.25}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
train-optimizer_batch-057 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 16}, "work": {"bytes_per_param": 8, "fixed_memory_gb": 16.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-two_oom_clipping-052 | two_oom_clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["OOM_BATCH_TOO_LARGE", "NO_GRADIENT_CLIPPING"], "fix": {"max_grad_norm": 1.0, "per_device_batch_size": 70}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-checkpoint_retention-055 | checkpoint_retention | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINT_RETENTION_LIMIT", "fix": {"max_checkpoints_to_keep": 1}, "work": {"checkpoint_size_gb": 112.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-fp16_scaler-006 | fp16_scaler | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "FP16_NO_GRAD_SCALER", "fix": {"use_grad_scaler": true}} | {"archetype": "fp16_scaler", "computations": ["R6"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-parallel_batch-045 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 25}, "work": {"fixed_memory_gb": 128.0, "params_per_rank_billions": 8.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-two_lora_vocab-035 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-effective_batch-001 | effective_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "EFFECTIVE_BATCH_MISMATCH", "fix": {"grad_accum_steps": 1}} | {"archetype": "effective_batch", "computations": ["R3"], "depth": 2, "interaction": null, "tier": "diagnosis"} |
train-optimizer_batch-061 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 24}, "work": {"bytes_per_param": 8, "fixed_memory_gb": 56.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-two_oom_clipping-019 | two_oom_clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["OOM_BATCH_TOO_LARGE", "NO_GRADIENT_CLIPPING"], "fix": {"max_grad_norm": 1.0, "per_device_batch_size": 5}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-packing_time-009 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 0.5}, "work": {"total_padded_tokens": 163840000, "total_real_tokens": 144000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-accum_and_steps-029 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 4, "planned_total_steps": 200}, "work": {"steps_per_epoch": 100}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-two_sampler_accum-038 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 4, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-lr_at_step-030 | lr_at_step | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "LR_VALUE_AT_STEP", "fix": {"current_learning_rate": 0.000155}, "work": {"step_fraction": 0.25}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
train-shard_size-035 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 10000, "shard_size_gb": 12.5}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-ckpt_batch_accum-062 | ckpt_batch_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINTED_BATCH_AND_ACCUM", "fix": {"grad_accum_steps": 4, "per_device_batch_size": 17}, "work": {"activation_per_sample_gb": 2.0, "max_per_device_batch": 17}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-parallel_batch-052 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 85}, "work": {"fixed_memory_gb": 64.0, "params_per_rank_billions": 4.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-accum_and_steps-059 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 8, "planned_total_steps": 500}, "work": {"steps_per_epoch": 250}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-parallel_batch-024 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 38}, "work": {"fixed_memory_gb": 32.0, "params_per_rank_billions": 2.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-lr_at_step-041 | lr_at_step | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "LR_VALUE_AT_STEP", "fix": {"current_learning_rate": 0.0002325}, "work": {"step_fraction": 0.25}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
train-optimizer_batch-023 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 4}, "work": {"bytes_per_param": 8, "fixed_memory_gb": 32.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-batch_from_components-026 | batch_from_components | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "BATCH_FROM_MEMORY_COMPONENTS", "fix": {"per_device_batch_size": 16}, "work": {"fixed_memory_gb": 64.0, "remaining_memory_gb": 16.0}} | {"archetype": "batch_from_components", "computations": ["R13", "R1"], "depth": 3, "interaction": null, "tier": "computed"} |
train-training_time-058 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.5688888888888889}, "work": {"total_tokens": 163840000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-lora_frozen-001 | lora_frozen | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ADAPTERS_FROZEN_ZERO_TRAINABLE", "fix": {"adapter_requires_grad": true}} | {"archetype": "lora_frozen", "computations": ["R8"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-checkpoint_retention-069 | checkpoint_retention | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINT_RETENTION_LIMIT", "fix": {"max_checkpoints_to_keep": 31}, "work": {"checkpoint_size_gb": 24.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-accum_steps_lr-030 | accum_steps_lr | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_STEPS_AND_LR", "fix": {"current_learning_rate": 6.5e-05, "grad_accum_steps": 8, "planned_total_steps": 200}, "work": {"audit_step": 150, "steps_per_epoch": 100}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-packing_time-006 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 0.5}, "work": {"total_padded_tokens": 163840000, "total_real_tokens": 144000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-ckpt_batch_accum-016 | ckpt_batch_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINTED_BATCH_AND_ACCUM", "fix": {"grad_accum_steps": 2, "per_device_batch_size": 49}, "work": {"activation_per_sample_gb": 1.5, "max_per_device_batch": 49}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-two_lora_vocab-036 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-accum_and_steps-015 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 2, "planned_total_steps": 1600}, "work": {"steps_per_epoch": 400}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-optimizer_batch-026 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 61}, "work": {"bytes_per_param": 12, "fixed_memory_gb": 36.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-zero_batch-010 | zero_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ZERO_SHARDED_BATCH_SIZE", "fix": {"per_device_batch_size": 39}, "work": {"bytes_per_param_per_rank": 7.25, "fixed_memory_gb": 29.0, "remaining_memory_gb": 99.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-zero_batch-024 | zero_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ZERO_SHARDED_BATCH_SIZE", "fix": {"per_device_batch_size": 12}, "work": {"bytes_per_param_per_rank": 4.0, "fixed_memory_gb": 16.0, "remaining_memory_gb": 24.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-batch_from_components-010 | batch_from_components | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "BATCH_FROM_MEMORY_COMPONENTS", "fix": {"per_device_batch_size": 6}, "work": {"fixed_memory_gb": 64.0, "remaining_memory_gb": 16.0}} | {"archetype": "batch_from_components", "computations": ["R13", "R1"], "depth": 3, "interaction": null, "tier": "computed"} |
train-two_oom_clipping-038 | two_oom_clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["OOM_BATCH_TOO_LARGE", "NO_GRADIENT_CLIPPING"], "fix": {"max_grad_norm": 1.0, "per_device_batch_size": 66}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-vocab-005 | vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "EMBEDDING_VOCAB_MISMATCH", "fix": {"embedding_vocab_size": 64032}} | {"archetype": "vocab", "computations": ["R11"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-training_time-035 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 2.2755555555555556}, "work": {"total_tokens": 491520000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-accum_and_steps-050 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 16, "planned_total_steps": 750}, "work": {"steps_per_epoch": 250}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-shard_size-044 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 10000, "shard_size_gb": 40.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-zero_batch-062 | zero_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ZERO_SHARDED_BATCH_SIZE", "fix": {"per_device_batch_size": 96}, "work": {"bytes_per_param_per_rank": 12.0, "fixed_memory_gb": 96.0, "remaining_memory_gb": 96.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-dataloader-005 | dataloader | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "DATALOADER_FORK_WITH_CUDA", "fix": {"dataloader_multiprocessing_context": "spawn"}} | {"archetype": "dataloader", "computations": ["R9"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-training_time-014 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.8533333333333334}, "work": {"total_tokens": 122880000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-two_lora_vocab-018 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-shard_size-037 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 12500, "shard_size_gb": 40.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-training_time-042 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.8533333333333334}, "work": {"total_tokens": 245760000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-two_sampler_accum-037 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 8, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-optimizer_batch-041 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 72}, "work": {"bytes_per_param": 12, "fixed_memory_gb": 24.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-two_lora_vocab-042 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-accum_and_steps-066 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 4, "planned_total_steps": 300}, "work": {"steps_per_epoch": 100}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-parallel_batch-051 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 80}, "work": {"fixed_memory_gb": 32.0, "params_per_rank_billions": 2.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-two_sampler_accum-051 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 2, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-two_lora_vocab-015 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 128256}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-shard_size-069 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 10000, "shard_size_gb": 12.5}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-training_time-059 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 2.1333333333333333}, "work": {"total_tokens": 307200000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-optimizer_batch-052 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 40}, "work": {"bytes_per_param": 12, "fixed_memory_gb": 48.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-two_v100_dataloader-014 | two_v100_dataloader | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["PRECISION_UNSUPPORTED_ON_GPU", "DATALOADER_FORK_WITH_CUDA"], "fix": {"dataloader_multiprocessing_context": "spawn", "dtype": "fp16", "use_grad_scaler": true}} | {"archetype": "two_v100_dataloader", "computations": ["R5", "R9"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-two_sampler_accum-041 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 8, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-training_time-049 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.4740740740740741}, "work": {"total_tokens": 163840000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-accum_steps_lr-034 | accum_steps_lr | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_STEPS_AND_LR", "fix": {"current_learning_rate": 0.000165, "grad_accum_steps": 8, "planned_total_steps": 500}, "work": {"audit_step": 250, "steps_per_epoch": 250}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-batch_from_components-065 | batch_from_components | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "BATCH_FROM_MEMORY_COMPONENTS", "fix": {"per_device_batch_size": 37}, "work": {"fixed_memory_gb": 24.0, "remaining_memory_gb": 56.0}} | {"archetype": "batch_from_components", "computations": ["R13", "R1"], "depth": 3, "interaction": null, "tier": "computed"} |
train-packing_time-004 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 0.5}, "work": {"total_padded_tokens": 81920000, "total_real_tokens": 72000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-ckpt_batch_accum-036 | ckpt_batch_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINTED_BATCH_AND_ACCUM", "fix": {"grad_accum_steps": 8, "per_device_batch_size": 20}, "work": {"activation_per_sample_gb": 1.5, "max_per_device_batch": 20}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-optimizer_batch-021 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 10}, "work": {"bytes_per_param": 10, "fixed_memory_gb": 70.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-accum_steps_lr-047 | accum_steps_lr | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_STEPS_AND_LR", "fix": {"current_learning_rate": 0.000195, "grad_accum_steps": 8, "planned_total_steps": 500}, "work": {"audit_step": 375, "steps_per_epoch": 250}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-accum_and_steps-055 | accum_and_steps | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACCUM_AND_TOTAL_STEPS", "fix": {"grad_accum_steps": 16, "planned_total_steps": 400}, "work": {"steps_per_epoch": 200}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-device_mismatch-002 | device_mismatch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "DEVICE_MISMATCH_LOSS", "fix": {"loss_device": "cuda:0"}} | {"archetype": "device_mismatch", "computations": ["R10"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-two_oom_clipping-002 | two_oom_clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["OOM_BATCH_TOO_LARGE", "NO_GRADIENT_CLIPPING"], "fix": {"max_grad_norm": 1.0, "per_device_batch_size": 12}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-two_sampler_accum-050 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 4, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-two_sampler_accum-043 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 4, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-two_v100_dataloader-020 | two_v100_dataloader | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["PRECISION_UNSUPPORTED_ON_GPU", "DATALOADER_FORK_WITH_CUDA"], "fix": {"dataloader_multiprocessing_context": "spawn", "dtype": "fp16", "use_grad_scaler": true}} | {"archetype": "two_v100_dataloader", "computations": ["R5", "R9"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-training_time-069 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.5688888888888889}, "work": {"total_tokens": 163840000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-optimizer_batch-060 | optimizer_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "OPTIMIZER_MEMORY_BATCH", "fix": {"per_device_batch_size": 5}, "work": {"bytes_per_param": 10, "fixed_memory_gb": 30.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-clipping-006 | clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "NO_GRADIENT_CLIPPING", "fix": {"max_grad_norm": 1.0}} | {"archetype": "clipping", "computations": ["R12"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-two_oom_clipping-040 | two_oom_clipping | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["OOM_BATCH_TOO_LARGE", "NO_GRADIENT_CLIPPING"], "fix": {"max_grad_norm": 1.0, "per_device_batch_size": 28}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-parallel_batch-044 | parallel_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PARALLEL_SHARDED_BATCH", "fix": {"per_device_batch_size": 48}, "work": {"fixed_memory_gb": 32.0, "params_per_rank_billions": 2.0}} | {"archetype": "parallel_batch", "computations": ["R22", "R13", "R1"], "depth": 4, "interaction": "parallelism x memory", "tier": "interacting"} |
train-two_sampler_accum-009 | two_sampler_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["DUPLICATE_DATA_NO_DISTRIBUTED_SAMPLER", "EFFECTIVE_BATCH_MISMATCH"], "fix": {"grad_accum_steps": 8, "sampler": "DistributedSampler"}} | {"archetype": "two_sampler_accum", "computations": ["R7", "R3"], "depth": 3, "interaction": "sampler+accum", "tier": "two_fault"} |
train-shard_size-062 | shard_size | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "SHARD_SIZE_PER_RANK", "fix": {"examples_per_rank": 12500, "shard_size_gb": 25.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-trap_oom_checkpointing-004 | trap_oom_checkpointing | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACTIVATION_MEMORY_NO_CHECKPOINTING", "fix": {"gradient_checkpointing": true}} | {"archetype": "trap_oom_checkpointing", "computations": ["R1", "R2"], "depth": 2, "interaction": "oom_batch", "tier": "trap"} |
train-two_lora_vocab-059 | two_lora_vocab | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": ["ADAPTERS_FROZEN_ZERO_TRAINABLE", "EMBEDDING_VOCAB_MISMATCH"], "fix": {"adapter_requires_grad": true, "embedding_vocab_size": 50304}} | {"archetype": "two_lora_vocab", "computations": ["R8", "R11"], "depth": 2, "interaction": "mixed", "tier": "two_fault"} |
train-training_time-019 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 0.4444444444444444}, "work": {"total_tokens": 153600000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-training_time-039 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 1.4222222222222223}, "work": {"total_tokens": 327680000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-training_time-004 | training_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "TRAINING_TIME_ESTIMATE", "fix": {"planned_time_hours": 2.1333333333333333}, "work": {"total_tokens": 245760000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-ckpt_batch_accum-013 | ckpt_batch_accum | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "CHECKPOINTED_BATCH_AND_ACCUM", "fix": {"grad_accum_steps": 8, "per_device_batch_size": 17}, "work": {"activation_per_sample_gb": 1.5, "max_per_device_batch": 17}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-zero_batch-051 | zero_batch | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ZERO_SHARDED_BATCH_SIZE", "fix": {"per_device_batch_size": 80}, "work": {"bytes_per_param_per_rank": 4.0, "fixed_memory_gb": 32.0, "remaining_memory_gb": 160.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-packing_time-060 | packing_time | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "PACKED_TRAINING_TIME", "fix": {"planned_time_hours": 1.0}, "work": {"total_padded_tokens": 327680000, "total_real_tokens": 144000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-trap_oom_checkpointing-010 | trap_oom_checkpointing | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "ACTIVATION_MEMORY_NO_CHECKPOINTING", "fix": {"gradient_checkpointing": true}} | {"archetype": "trap_oom_checkpointing", "computations": ["R1", "R2"], "depth": 2, "interaction": "oom_batch", "tier": "trap"} |
train-lr_at_step-037 | lr_at_step | train | You are debugging a failed or misbehaving LLM training run.
Pick exactly ONE diagnosis code and give the minimal fix: only the config fields that must change, with corrected values.
Diagnosis codes:
- OOM_BATCH_TOO_LARGE
- ACTIVATION_MEMORY_NO_CHECKPOINTING
- EFFECTIVE_BATCH_MISMATCH
- LR_TOO_HIGH_NO_WARMUP
- PRECISIO... | {"diagnosis": "LR_VALUE_AT_STEP", "fix": {"current_learning_rate": 0.0002325}, "work": {"step_fraction": 0.25}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.