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-two_sampler_accum-000 | 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-shard_size-059 | 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": 60.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-optimizer_batch-069 | 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-parallel_batch-067 | 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-ckpt_batch_accum-056 | 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": 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-040 | 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": 48}, "work": {"bytes_per_param_per_rank": 8.0, "fixed_memory_gb": 32.0, "remaining_memory_gb": 48.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-shard_size-012 | 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": 5000, "shard_size_gb": 60.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-ckpt_batch_accum-022 | 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": 35}, "work": {"activation_per_sample_gb": 2.0, "max_per_device_batch": 35}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-clipping-002 | 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-lr_at_step-026 | 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": 0.000165}, "work": {"step_fraction": 0.5}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
train-ckpt_batch_accum-050 | 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": 46}, "work": {"activation_per_sample_gb": 1.5, "max_per_device_batch": 46}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-lr_at_step-014 | 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-accum_steps_lr-045 | 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": 0.000155, "grad_accum_steps": 8, "planned_total_steps": 600}, "work": {"audit_step": 150, "steps_per_epoch": 200}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-training_time-053 | 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": 0.7585185185185185}, "work": {"total_tokens": 327680000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-parallel_batch-021 | 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": 51}, "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-zero_batch-000 | 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": 38}, "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-batch_from_components-043 | 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": 28}, "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-zero_batch-035 | 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": 76}, "work": {"bytes_per_param_per_rank": 7.25, "fixed_memory_gb": 116.0, "remaining_memory_gb": 76.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-trap_oom_checkpointing-003 | 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-038 | 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-checkpoint_retention-009 | 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": 8}, "work": {"checkpoint_size_gb": 112.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-checkpoint_retention-019 | 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": 8}, "work": {"checkpoint_size_gb": 24.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-packing_time-051 | 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-two_sampler_accum-044 | 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-ckpt_batch_accum-058 | 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": 66}, "work": {"activation_per_sample_gb": 1.0, "max_per_device_batch": 66}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-checkpoint_retention-016 | 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": 15}, "work": {"checkpoint_size_gb": 48.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-lora_frozen-006 | 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-two_lora_vocab-034 | 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-bf16_v100-003 | bf16_v100 | 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", "fix": {"dtype": "fp16", "use_grad_scaler": true}} | {"archetype": "bf16_v100", "computations": ["R5", "R6"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-lora_frozen-004 | 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-parallel_batch-036 | 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": 10}, "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-020 | 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": 800}, "work": {"steps_per_epoch": 200}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-optimizer_batch-054 | 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": 104}, "work": {"bytes_per_param": 8, "fixed_memory_gb": 24.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-accum_and_steps-056 | 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": 800}, "work": {"steps_per_epoch": 200}} | {"archetype": "accum_and_steps", "computations": ["R3", "R14"], "depth": 3, "interaction": null, "tier": "computed"} |
train-packing_time-039 | 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-two_lora_vocab-044 | 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-lr_at_step-063 | 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-two_v100_dataloader-008 | 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-dataloader-001 | 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-checkpoint_retention-042 | 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": 8}, "work": {"checkpoint_size_gb": 112.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-two_v100_dataloader-058 | 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-clipping-000 | 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-effective_batch-004 | 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": 32}} | {"archetype": "effective_batch", "computations": ["R3"], "depth": 2, "interaction": null, "tier": "diagnosis"} |
train-trap_oom_checkpointing-000 | 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-049 | 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-two_v100_dataloader-049 | 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_v100_dataloader-042 | 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_oom_clipping-012 | 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-two_oom_clipping-023 | 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": 7}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-training_time-009 | 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": 245760000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-zero_batch-031 | 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": 176}, "work": {"bytes_per_param_per_rank": 2.0, "fixed_memory_gb": 16.0, "remaining_memory_gb": 176.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-checkpointing-004 | 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": "checkpointing", "computations": ["R2"], "depth": 1, "interaction": null, "tier": "diagnosis"} |
train-training_time-018 | 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": 1.517037037037037}, "work": {"total_tokens": 327680000}} | {"archetype": "training_time", "computations": ["R16"], "depth": 3, "interaction": null, "tier": "computed"} |
train-ckpt_batch_accum-068 | 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": 30}, "work": {"activation_per_sample_gb": 1.0, "max_per_device_batch": 30}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-two_lora_vocab-026 | 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-two_oom_clipping-005 | 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": 30}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-optimizer_batch-020 | 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": 48}, "work": {"bytes_per_param": 8, "fixed_memory_gb": 24.0}} | {"archetype": "optimizer_batch", "computations": ["R21", "R1"], "depth": 3, "interaction": "optimizer x memory", "tier": "interacting"} |
train-two_v100_dataloader-036 | 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-zero_batch-030 | 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": 38}, "work": {"bytes_per_param_per_rank": 2.0, "fixed_memory_gb": 32.0, "remaining_memory_gb": 96.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-shard_size-052 | 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": 5000, "shard_size_gb": 25.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-batch_from_components-053 | 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": 21}, "work": {"fixed_memory_gb": 48.0, "remaining_memory_gb": 32.0}} | {"archetype": "batch_from_components", "computations": ["R13", "R1"], "depth": 3, "interaction": null, "tier": "computed"} |
train-packing_time-023 | 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": 327680000, "total_real_tokens": 144000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-packing_time-037 | 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": 216000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-lr_at_step-066 | 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": 0.00011}, "work": {"step_fraction": 0.5}} | {"archetype": "lr_at_step", "computations": ["R15"], "depth": 2, "interaction": null, "tier": "computed"} |
train-checkpoint_retention-047 | 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": 8}, "work": {"checkpoint_size_gb": 112.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-zero_batch-007 | 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": 56}, "work": {"bytes_per_param_per_rank": 4.0, "fixed_memory_gb": 16.0, "remaining_memory_gb": 112.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-two_sampler_accum-049 | 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-two_v100_dataloader-057 | 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-shard_size-049 | 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": 5000, "shard_size_gb": 60.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-two_v100_dataloader-004 | 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-019 | 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": 122880000, "total_real_tokens": 108000000}} | {"archetype": "packing_time", "computations": ["R23", "R16"], "depth": 4, "interaction": "packing x time", "tier": "interacting"} |
train-two_sampler_accum-005 | 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-057 | 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-parallel_batch-055 | 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-ckpt_batch_accum-023 | 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": 44}, "work": {"activation_per_sample_gb": 1.5, "max_per_device_batch": 44}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-lr_at_step-025 | 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-parallel_batch-041 | 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": 32}, "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-011 | 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-011 | 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-ckpt_batch_accum-009 | 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": 13}, "work": {"activation_per_sample_gb": 2.0, "max_per_device_batch": 13}} | {"archetype": "ckpt_batch_accum", "computations": ["R20", "R1", "R3"], "depth": 4, "interaction": "checkpointing x accum", "tier": "interacting"} |
train-oom_batch-006 | oom_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": "OOM_BATCH_TOO_LARGE", "fix": {"per_device_batch_size": 23}} | {"archetype": "oom_batch", "computations": ["R1"], "depth": 2, "interaction": null, "tier": "diagnosis"} |
train-two_oom_clipping-001 | 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": 37}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-two_oom_clipping-021 | 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-device_mismatch-004 | 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-accum_steps_lr-006 | 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": 0.0002325, "grad_accum_steps": 2, "planned_total_steps": 200}, "work": {"audit_step": 50, "steps_per_epoch": 100}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-accum_steps_lr-007 | 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": 16, "planned_total_steps": 750}, "work": {"audit_step": 562, "steps_per_epoch": 250}} | {"archetype": "accum_steps_lr", "computations": ["R3", "R14", "R15"], "depth": 5, "interaction": "accum x schedule", "tier": "interacting"} |
train-zero_batch-063 | 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": 78}, "work": {"bytes_per_param_per_rank": 9.0, "fixed_memory_gb": 36.0, "remaining_memory_gb": 156.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-parallel_batch-037 | 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": 21}, "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-trap_nan_clipping-011 | trap_nan_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": "trap_nan_clipping", "computations": ["R4", "R12"], "depth": 2, "interaction": "lr", "tier": "trap"} |
train-parallel_batch-012 | 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-lr_at_step-059 | 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-two_oom_clipping-046 | 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": 9}} | {"archetype": "two_oom_clipping", "computations": ["R1", "R12"], "depth": 3, "interaction": "oom+clipping", "tier": "two_fault"} |
train-zero_batch-058 | 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": 42}, "work": {"bytes_per_param_per_rank": 2.0, "fixed_memory_gb": 16.0, "remaining_memory_gb": 64.0}} | {"archetype": "zero_batch", "computations": ["R19", "R13", "R1"], "depth": 4, "interaction": "zero x memory", "tier": "interacting"} |
train-shard_size-005 | 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": 5000, "shard_size_gb": 25.0}} | {"archetype": "shard_size", "computations": ["R18"], "depth": 2, "interaction": null, "tier": "computed"} |
train-two_v100_dataloader-044 | 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-checkpoint_retention-026 | 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": 6}, "work": {"checkpoint_size_gb": 112.0}} | {"archetype": "checkpoint_retention", "computations": ["R17"], "depth": 2, "interaction": null, "tier": "computed"} |
train-batch_from_components-040 | 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_v100_dataloader-040 | 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-parallel_batch-032 | 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": 12}, "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-batch_from_components-047 | 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": 4}, "work": {"fixed_memory_gb": 32.0, "remaining_memory_gb": 8.0}} | {"archetype": "batch_from_components", "computations": ["R13", "R1"], "depth": 3, "interaction": null, "tier": "computed"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.