Downstream Validation Using Eleuther Eval Harness

This page walks through the steps for performing downstream validation on the Cerebras Wafer-Scale cluster using EleutherAI’s Evaluation Harness (EEH). EEH is a popular framework for evaluating large language models across various different datasets and tasks.

While you can configure EEH as part of your training (see Pretraining with Downstream Validation from the Cerebras Model Zoo.

The examples in this guide will perform downstream validation on LLaMA3 8B.

By the end of this guide, you will be able to leverage the EEH framework to perform standalone downstream validation on your models on CS-X.

To skip ahead to the examples covered in this guide without following the entire walk through, please follow the links:

Example (Single Non-generative Task)

Let’s assume that the YAML configuration file above is written to ./llama3_8B_eeh.yaml. Then, to run evaluation for task winogrande, please set up a bash script as follows:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
  --params ./llama3_8B_eeh.yaml \
  --checkpoint_path <path_to_checkpoint_file> \
  --python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
  --mount_dirs <path(s)_to_mount_to_appliance_containers> \ # Must include the path to the "data_dir" eval harness config in the YAML

The output logs are as follows:

...
2024-06-24 11:06:49,596 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=632, Rate=34.38 samples/sec, GlobalRate=34.34 samples/sec
2024-06-24 11:06:49,625 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=633, Rate=34.39 samples/sec, GlobalRate=34.35 samples/sec
2024-06-24 11:07:04,333 INFO:   <cerebras.modelzoo.trainer.extensions.eleuther.lm_eval_harness.EleutherLM object at 0x7ff89ae3ceb0> (None), gen_kwargs: ({'temperature': None, 'top_k': None, 'top_p': None, 'max_tokens': None}), limit: None, num_fewshot: None, batch_size: None
2024-06-24 11:07:04,696 INFO:
|  Tasks   |Version|Filter|n-shot|Metric|Value |   |Stderr|
|----------|------:|------|-----:|------|-----:|---|-----:|
|winogrande|      1|none  |     0|acc   |0.7284|±  |0.0141|

Example (Multiple Non-generative Tasks)

To run evaluation on more non-generative tasks, you may update the run script to the following:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
  --params ./llama3_8B_eeh.yaml \
  --tasks "arc_challenge,hellaswag,openbookqa,piqa,winogrande" \ # Overrides the EEH "tasks" config in the YAML
  --checkpoint_path <path_to_checkpoint_file> \
  --python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
  --mount_dirs <path(s)_to_mount_to_appliance_containers> \

The output logs are as follows:

...
2024-06-24 12:40:44,896 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=2979, Rate=34.40 samples/sec, GlobalRate=34.39 samples/sec
2024-06-24 12:40:57,106 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=2980, Rate=34.40 samples/sec, GlobalRate=34.39 samples/sec
2024-06-24 12:41:10,333 INFO:   <cerebras.modelzoo.trainer.extensions.eleuther.lm_eval_harness.EleutherLM object at 0x7ff89ae3ceb0> (None), gen_kwargs: ({'temperature': None, 'top_k': None, 'top_p': None, 'max_tokens': None}), limit: None, num_fewshot: None, batch_size: None
2024-06-24 12:41:10,700 INFO:
|    Tasks    |Version|Filter|n-shot| Metric |Value |   |Stderr|
|-------------|-------|------|-----:|--------|-----:|---|-----:|
|arc_challenge|Yaml   |none  |     0|acc     |0.5334|±  |0.0145|
|             |       |none  |     0|acc_norm|0.5300|±  |0.0146|
|hellaswag    |Yaml   |none  |     0|acc     |0.5716|±  |0.0049|
|             |       |none  |     0|acc_norm|0.7917|±  |0.0043|
|openbookqa   |Yaml   |none  |     0|acc     |0.3210|±  |0.0207|
|             |       |none  |     0|acc_norm|0.4520|±  |0.0223|
|piqa         |Yaml   |none  |     0|acc     |0.7997|±  |0.0097|
|             |       |none  |     0|acc_norm|0.8090|±  |0.0095|
|winogrande   |Yaml   |none  |     0|acc     |0.7284|±  |0.0130|

Example (Generative task)

The EEH flow also supports running generative (autoregressive) eval harness tasks, i.e. tasks that specify output_type: generate_until in the task specification, such as triviaqa or drop. Refer to task triviqa for an example specification from the official EEH repository.

In order to run generative inference on CSX, you must specify the following inference settings in the model config of YAML file:

• start_token - ID of the special token that indicates where to start inferring for each sample, as described above. You may specify a list of token IDs instead of a single ID. If you do, the model will start inference at the first token that matches any one of the provided IDs. The model will pad inferred predictions with the first ID in the list.

• stop_sequences - List of sequences (each one being a list of token IDs). If any one of these sequences is emitted by the model, inference will stop for that sample. For example, suppose you would like to stop inferring after either a newline character (e.g. token id 1), or a combination of a period (e.g. token id 2) followed by a space (e.g. token id 3). In this case, set stop_sequences to [[1], [2, 3]]. To stop inferring after seeing a newline character only, set stop_sequences to [[1]]. To disable this feature, set stop_sequences to an empty list []. Additionally, the following optional parameters may be set:

• max_tokens - Maximum tokens to infer for each sample.

• loop_dim - Indicates the sequence dimension in the input and output data. Default value is 1. If set to 0, indicates that both input and output data is transposed (i.e. sequence X samples instead of samples X sequence).

For your LLaMA3 8B example, please update the YAML file ./llama3_8B_eeh.yaml to add these inference settings:

trainer:
  init:
    backend: # CSX
      ...
    model: # Llama3-8B
      name: llama
      ...
      # Inference Settings
      start_token: 128256                 # Set to `vocab_size`
      stop_sequences: [[198], [13], [11]] # Respective tokens for "\n", "." and ","
      max_tokens: 256                     # Default from HF implementations
      loop_dim: 1
    callbacks:
    - EleutherEvalHarness:
        # Eleuther Eval Harness settings
        eeh_args:
          tasks: triviaqa
          num_fewshot: 0
      ...
  ...

1. For start_token, it is ideal to choose a value that’s not going to be generated by the model, i.e. vocab_size in the example above.

2. The generative task itself defines stop_sequences under setting generation_kwargs.until of the task spec. For instance, triviqa specifies "\n", "." and "," as the stop tokens. The EEH flow will internally override the stop_sequences config with the value from the task, so you can also specify an arbitrary value in the YAML.

Finally, please update the bash script as follows:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
  --params ./llama3_8B_eeh.yaml \
  --checkpoint_path <path_to_checkpoint_file> \
  --python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
  --mount_dirs <path(s)_to_mount_to_appliance_containers> \

The output logs are as follows:

...
2024-06-26 11:22:47,530 INFO:   | EleutherAI Generative Eval Device=CSX, GlobalStep=1, Batch=4480, Rate=0.38 samples/sec, GlobalRate=0.38 samples/sec
2024-06-26 11:22:47,553 INFO:   | EleutherAI Generative Eval Device=CSX, GlobalStep=1, Batch=4485, Rate=0.38 samples/sec, GlobalRate=0.38 samples/sec
2024-06-26 11:38:15,868 INFO:   <cerebras.modelzoo.trainer.extensions.eleuther.lm_eval_harness.EleutherLM object at 0x7f9e305fa730> (None), gen_kwargs: ({'temperature': None, 'top_k': None, 'top_p': None, 'max_tokens': None}), limit: None, num_fewshot: None, batch_size: None
2024-06-26 11:38:16,232 INFO:
| Tasks  |Version|     Filter      |n-shot|  Metric   |Value |   |Stderr|
|--------|------:|-----------------|-----:|-----------|-----:|---|-----:|
|triviaqa|      3|remove_whitespace|     0|exact_match|0.1120|±  |0.0001|

By default, the model will perform greedy sampling of the inferred tokens, i.e. for all of the model’s outputs, pick the token with the highest probability.

In order to perform non-greedy sampling, you can pass in temperature, top_k or top_p to either the bash script or under eeh_args of the YAML. For example:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
  --params ./llama3_8B_eeh.yaml \
  --checkpoint_path <path_to_checkpoint_file> \
  --python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
  --mount_dirs <path(s)_to_mount_to_appliance_containers> \
  --temperature 0.7 \
  --top_k 10 \
  --top_p 0.95 \

Example (Non-generative and Generative Tasks)

You can combine evaluation for generative and non-generative eval harness tasks simply via updating the bash script as follows:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
  --params ./llama3_8B_eeh.yaml \
  --tasks winogrande,drop \ # Runs evaluation for non-generative task "winogrande" and generative task "drop"
  --checkpoint_path <path_to_checkpoint_file> \
  --python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
  --mount_dirs <path(s)_to_mount_to_appliance_containers> \

The output logs are as follows:

...
2024-06-26 13:57:52,177 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=632, Rate=34.38 samples/sec, GlobalRate=34.34 samples/sec
2024-06-26 13:57:52,206 INFO:   | EleutherAI Eval Device=CSX, GlobalStep=1, Batch=633, Rate=34.39 samples/sec, GlobalRate=34.35 samples/sec
2024-06-26 13:57:52,333 INFO:   Running generate_until requests
...
2024-06-26 14:00:03,499 INFO:   | EleutherAI Generative Eval Device=CSX, GlobalStep=1, Batch=2382, Rate=0.38 samples/sec, GlobalRate=0.38 samples/sec
2024-06-26 14:00:03,519 INFO:   | EleutherAI Generative Eval Device=CSX, GlobalStep=1, Batch=2383, Rate=0.38 samples/sec, GlobalRate=0.38 samples/sec
2024-06-26 14:00:31,890 INFO:   <cerebras.modelzoo.trainer.extensions.eleuther.lm_eval_harness.EleutherLM object at 0x7f156522c490> (None), gen_kwargs: ({'temperature': None, 'top_k': None, 'top_p': None, 'max_tokens': None}), limit: None, num_fewshot: None, batch_size: None
2024-06-26 14:00:32,351 INFO:
|  Tasks   |Version|Filter|n-shot|Metric|Value |   |Stderr|
|----------|------:|------|-----:|------|-----:|---|-----:|
|winogrande|      1|none  |     0|acc   |0.7284|±  |0.0141|
|drop      |      3|none  |     0|em    |0.0072|±  |0.0008|
|          |       |none  |     0|f1    |0.0371|±  |0.0013|

1. Since inference settings are baked into the model configuration and that inference requires different resources, there is a separate compile and execution for running downstream validation on generative tasks.

2. You may specify at most one generative task at a time (per callback).

Run Multiple Generative Tasks

In order to run multiple generative tasks, you may update the run script to the following:

python <path_to_cerebras_modelzoo>/common/run_eleuther_eval_harness.py CSX \
--params ./llama3_8B_eeh.yaml \
--tasks "drop,gsm8k" \ # Overrides the EEH "tasks" config in the YAML
--checkpoint_path <path_to_checkpoint_file> \
--python_paths <path(s)_to_export_to_PYTHONPATH_in_appliance_containers> \
--mount_dirs <path(s)_to_mount_to_appliance_containers> \ 

Prerequisites

Please ensure that you have installed the Cerebras Model Zoo package by going through the installation guide. Note that EEH version tested and packaged in the Cerebras Model Zoo is the official release v0.4.2.

Please also read through the Trainer Overview and Trainer Configuration Overview, as these guides will help understand how to configure running EEH standalone.

This guide configures the downstream EEH run using V2 YAML. While release 2.3 includes support for legacy V1 YAML, please convert the configuration to V2 using convert_legacy_params_to_trainer_params from the script src/cerebras/modelzoo/trainer/utils.py.

Configure the Run

This section covers the required steps for setting up an EEH run to perform standalone downstream validation on various tasks.

In particular, you will need to write a YAML configuration file to configure an instance of the Trainer callback.

The example in this section configures evaluation for LLaMA3 8B via the multiple choice (non-generative) eval harness task winogrande using a single CSX.

If you aren’t interested in seeing the break down of the configuration, feel free to skip ahead to the Putting it All Together section to see the full YAML configuration.

Configure the CSX Backend

The first step is to specify the CSX backend and resources required for the run.

Please create a YAML configuration file with the following cluster config:

trainer:
  init:
    backend:
      backend_type: CSX
      cluster_config:
        num_csx: 1

This example uses a single CSX, but you can readily update num_csx to run EEH on multiple CSXs for improved performance.

Configure the Model

Next, please add the following model configuration in the YAML for LLaMA3 8B with 8K context length:

trainer:
  init:
    backend:  # CSX
      ...
    model:
      name: llama # This setting is required

      # Embedding
      vocab_size: 128256
      hidden_size: 4096
      position_embedding_type: "rotary"
      pos_scaling_factor: 1.0
      rope_theta: 500000.0
      rotary_dim: 128
      share_embedding_weights: false
      max_position_embeddings: 8192
      embedding_dropout_rate: 0.0
      embedding_layer_norm: false

      # Decoder
      num_hidden_layers: 32
      dropout_rate: 0.0
      layer_norm_epsilon: 1.0e-5
      norm_type: "rmsnorm"

      # Decoder - Attention
      num_heads: 32
      attention_type: "scaled_dot_product"
      attention_module: "multiquery_attention"
      attention_dropout_rate: 0.0
      use_projection_bias_in_attention: false
      use_ffn_bias_in_attention: false
      extra_attention_params:
        num_kv_groups: 8

      # Decoder - ffn
      filter_size: 14336
      nonlinearity: "swiglu"
      use_ffn_bias: false

      # Task-specific
      use_bias_in_output: false
      loss_scaling: "num_tokens"
      loss_weight: 1.0

      # Initializer
      initializer_range: 0.02

To run downstream validation harness, you must specify the name setting in the model configuration. Valid names corresponding to the supported models include:

• btlm

• bloom

• gpt2

• gptj

• falcon

• gpt3

• gpt-neox

• llama

• mistral

• mpt

• jais

• santacoder

• starcoder

Configure the EEH Callback

EEH is implemented as an extension to the Trainer callback.

Add the following section in the YAML to set up the EleutherEvalHarness callback:

trainer:
  init:
    backend: # CSX
      ...
    model: # Llama3-8B
      name: llama # This setting is required
      ...
    callbacks:
    - EleutherEvalHarness:
        # Eleuther Eval Harness settings (also exposed via CLI)
        eeh_args:
          tasks: winogrande
          num_fewshot: 0
        # CSX-specific eval harness settings (also exposed via CLI)
        keep_data_dir: false
        # Dataloader settings
        batch_size: 4
        shuffle: false
        max_sequence_length: 8192
        num_workers: 1
        data_dir: <path_to_mounted_dir>
        tokenizer_file_path: <path_to_llama3_tokenizer_json_file>
        eos_id: 128001
        pretrained_model_name_or_path: null
        # Eval Harness Flags
        flags:
          csx.performance.micro_batch_size: null

The eeh_args section exposes the following settings to configure the EEH run:

Eleuther Eval Harness CLI Arguments	Description
--tasks	Comma separated string specifying Eleuther Eval Harness tasks. To get full list of tasks, use the command lm-eval --tasks list from within your python venv.
--num_fewshot	Number of examples to be added to the fewshot context string. Defaults to 0
--output_path	The path to the output file where the result metrics will be saved. If the path is a directory and log_samples is true, the results will be saved in the directory. Else the parent directory will be used.
--limit	Accepts an integer, or a float between 0.0 and 1.0. This limits the number of documents to evaluate per task to the first X documents (if an integer) or first X% of documents. This is useful for debugging.
--use_cache	A path to a sqlite db file for caching model responses. None if not caching.
--cache_requests {true,refresh,delete}	Speed up evaluation by caching the building of dataset requests. None if not caching.
--check_integrity	Whether to run the relevant part of the test suite for the tasks.
--write_out	Prints the prompt for the first few documents. Defaults to False.
--log_samples	If True, write out all model outputs and documents for per-sample measurement and post-hoc analysis. Defaults to False.
--show_config	If True, shows the the full config of all tasks at the end of the evaluation. Defaults to False.
--include_path	Additional path to include if there are external tasks to include.
--predict_only	Use with –log_samples. Only model outputs will be saved and metrics will not be evaluated.
--seed	Set seed for python’s random, numpy and torch.
--temperature	Sampling temperature used for generation (autoregressive, generate_until tasks only).
--top_p	Top-p parameter used for nucleus sampling (autoregressive, generate_until tasks only).
--top_k	Top-k parameter used for generation (autoregressive, generate_until tasks only).

You can either specify the settings here or pass them via CLI arguments to the standalone EEH run script.

The callback configuration also accepts dataloader settings that you must specify in the YAML to set up input data preprocessing for the run:

DataLoader Settings	Description
data_dir	This setting is required. Provide a path to the mounted directory visible to the worker containers where eval harness task data samples are dumped after preprocessing. Use the mount_dirs argument to specify a dir mount, similar to our existing flows.
tokenizer_file_path	Path to a custom tokenizer (JSON) file. If you provide a custom tokenizer, then you must also specify eos_id; otherwise, you must provide a pretrained tokenizer from Hugging Face in pretrained_model_name_or_path.
pretrained_model_name_or_path	Hugging Face (HF) pretrained model name or path. This setting is required if you do not specify tokenizer_file_path. For detailed description, see HF AutoTokenizers.
eos_id	End-of-sentence (eos) token ID to signal the termination of a sequence. This setting is required if you specify a custom tokenizer in tokenizer_file_path. You can set this by looking for the ID corresponding to the eos token in the custom tokenizer JSON file.
max_sequence_length	Maximum length of the input sequence. This setting is required for preprocessing input data samples from the specified eval harness tasks. You should align the max_sequence_length field to the max_position_embeddings value in the model configuration of the YAML. If you don’t specify max_sequence_length, the flow defaults to this max_position_embeddings setting.

Additionally, you may optionally specify the following, CSX-specific eval harness setting:

• keep_data_dir: Use this to preserve the preprocessed eval harness task data samples, i.e. the directory specified under data_dir. Defaults to False, i.e. data samples are deleted after the run.

(Optional) Configure HuggingFace (HF) Cache Directory

EEH utilizes HF’s APIs to download task data and other configurations. This data is by default cached under $HOME/.cache/huggingface.

However, you may choose to specify a different directory for this cached data via the HFCacheDir callback:

trainer:
  init:
    backend: # CSX
      ...
    model: # Llama3-8B
      ...
    callbacks:
    - EleutherEvalHarness:
        ...
    - HFCacheDir:
        cache_dir: <path_to_directory_for_caching_HF_data>

Putting it All Together

Here’s what the full YAML configuration looks like once you follow this guide for configuring the individual pieces:

ltrainer:
init:
  backend:
    backend_type: CSX
    cluster_config:
      num_csx: 1
      mount_dirs: <path(s)_to_mount_to_appliance_containers>
  model:
    name: llama

    # Embedding
    vocab_size: 128256
    hidden_size: 4096
    position_embedding_type: "rotary"
    pos_scaling_factor: 1.0
    rope_theta: 500000.0
    rotary_dim: 128
    share_embedding_weights: false
    max_position_embeddings: 8192
    embedding_dropout_rate: 0.0
    embedding_layer_norm: false

    # Decoder
    num_hidden_layers: 32
    dropout_rate: 0.0
    layer_norm_epsilon: 1.0e-5
    norm_type: "rmsnorm"

    # Decoder - Attention
    num_heads: 32
    attention_type: "scaled_dot_product"
    attention_module: "multiquery_attention"
    attention_dropout_rate: 0.0
    use_projection_bias_in_attention: false
    use_ffn_bias_in_attention: false
    extra_attention_params:
      num_kv_groups: 8

    # Decoder - ffn
    filter_size: 14336
    nonlinearity: "swiglu"
    use_ffn_bias: false

    # Task-specific
    use_bias_in_output: false
    loss_scaling: "num_tokens"
    loss_weight: 1.0

    # Initializer
    initializer_range: 0.02
  callbacks:
  - EleutherEvalHarness:
      eeh_args:
        tasks: winogrande
        num_fewshot: 0
      keep_data_dir: false
      # Dataloader settings
      batch_size: 4
      shuffle: false
      max_sequence_length: 8192
      num_workers: 1
      data_dir: <path_to_mounted_dir>
      tokenizer_file_path: <path_to_llama3_tokenizer_json_file>
      eos_id: 128001
      pretrained_model_name_or_path: null
      # Eval Harness Flags
      flags:
        csx.performance.micro_batch_size: null

Running EEH on CS-X

Now that the the YAML configuration is complete, you will use the run script src/cerebras/modelzoo/common/run_eleuther_eval_harness.py from the Cerebras Model Zoo to run EEH on various tasks.

This script accepts the following command line interface (CLI) arguments:

python run_eleuther_eval_harness.py CSX [-h] [--tasks task1,task2] [--num_fewshot N] [--output_path DIR|DIR/file.json] [--limit N|0<N<1] [--use_cache DIR]
                                  [--cache_requests {true,refresh,delete}] [--check_integrity] [--write_out] [--log_samples] [--show_config] [--include_path DIR]
                                  [--predict_only] [--seed SEED] [--trust_remote_code] [--temperature TEMPERATURE] [--top_p TOP_P] [--top_k TOP_K]
                                  [--keep_data_dir]
                                  -p PARAMS [-m {eval}] [-o MODEL_DIR] [--checkpoint_path CHECKPOINT_PATH]
                                  [--disable_strict_checkpoint_loading] [--load_checkpoint_states LOAD_CHECKPOINT_STATES] [--logging LOGGING]
                                  [--wsc_log_level WSC_LOG_LEVEL [WSC_LOG_LEVEL ...]] [--max_steps MAX_STEPS] [--eval_steps EVAL_STEPS] [--config CONFIG]
                                  [--compile_only | --validate_only] [--num_workers_per_csx NUM_WORKERS_PER_CSX] [-c COMPILE_DIR]
                                  [--job_labels JOB_LABELS [JOB_LABELS ...]] [--job_priority {p1,p2,p3}] [--debug_args_path DEBUG_ARGS_PATH]
                                  [--mount_dirs MOUNT_DIRS [MOUNT_DIRS ...]] [--python_paths PYTHON_PATHS [PYTHON_PATHS ...]]
                                  [--credentials_path CREDENTIALS_PATH] [--mgmt_address MGMT_ADDRESS] [--job_time_sec JOB_TIME_SEC] [--disable_version_check]
                                  [--num_csx NUM_CSX] [--num_wgt_servers NUM_WGT_SERVERS] [--num_act_servers NUM_ACT_SERVERS]
                                  [--debug_args [DEBUG_ARGS [DEBUG_ARGS ...]]] [--ini [INI [INI ...]]] [--transfer_processes TRANSFER_PROCESSES]

1. We support a subset of Eleuther’s command line interface (CLI) arguments above. For a more detailed descrition of these supported arguments, see https://github.com/EleutherAI/lm-evaluation-harness/blob/v0.4.0/docs/interface.md.

2. You may also specify these arguments in the YAML under the eeh_args key of the EleutherEvalHarness configuration, but please note that the CLI setting will override the settings in the YAML.

3. The --params CLI argument is required. Use it to specify the path to the YAML configuration file. Note that while we do support the old V1 YAML specification, it will soon be deprecated so we recommend using the new V2 YAML.

4. Use the --checkpoint_path CLI argument to specify the path to the checkpoint file to load model weights from. If a checkpoint path is not provided, we support checkpoint autoloading in this flow such that the latest checkpoint file will be picked up from the specified model_dir.

Supported Tasks

• As of our 2.4.0 release, we support lm_eval@v0.4.5 .

• You may perform downstream validation on all EEH tasks with output_type: loglikelihood or output_type: multiple_choice in the task specification. See asdiv and arc_easy for respective examples. You may specify each of these types of tasks separately or together in a single EleutherEvalHarness callback.

• We currently do not support eval harness tasks with output_type: loglikelihood_rolling. These unsupported tasks are pile and wikitext for the supported EEH version 0.4.2.

  • agieval

  • babi

  • bbh

  • bbh_cot_fewshot

  • bbh_cot_zeroshot

  • bbh_fewshot

  • bbh_zeroshot

  • bigbench

  • codexglue_code2text

  • math_word_problems

  • mgsm_direct

  • mgsm_cot_native

  • mmlu_flan_cot_fewshot

  • mmlu_flan_cot_zeroshot

  • mmlu_flan_n_shot_generative

  • polemo2

  • super-glue-lm-eval-v1

  • unscramble

Adding New Tasks

Please refer to Eleuther’s new task implementation guide here to add new tasks.

Limitations

• We currently do not support running multiple generative eval harness tasks in the same callback.

• EEH task groups, such as agieval, comprise multiple generative sub tasks that you will have to configure in the YAML via separate callbacks.

• Please turn on grad accumulation and choose a small micro batch size (between 16 to 32) under the flags configuration of the EleutherEvalHarness callback of the YAML,

Conclusion

In summary, by following this guide you have run standalone downstream validation for the Llama3-8B model on EEH’s multiple different evaluation datasets / tasks.

You should now be comfortable in configuring more downstream EEH runs on your model of choice on even more eval harness tasks.

What’s next?

To run downstream validation on code generation tasks, please see check out:

• Downstream Validation using BigCode Eval Harness

You can also perform downstream validation using EEH as part of your pretraining runs with upstream validation. Check out the following guide:

• Pretraining with Downstream Validation