metadata
dataset_info:
features:
- name: request_id
dtype: string
- name: received_at
dtype: string
- name: completed_at
dtype: string
- name: caller_service
dtype: string
- name: inferred_caller
dtype: string
- name: llm_source
dtype: string
- name: model
dtype: string
- name: endpoint_kind
dtype: string
- name: path
dtype: string
- name: status_code
dtype: int64
- name: ok
dtype: bool
- name: stream
dtype: bool
- name: finish_reason
dtype: string
- name: system_prompt
dtype: string
- name: instruction
dtype: string
- name: input
dtype: string
- name: user_message_full
dtype: string
- name: output
dtype: string
- name: prompt_template
dtype: string
- name: split_marker
dtype: string
- name: prompt_tokens
dtype: int64
- name: completion_tokens
dtype: int64
- name: total_tokens
dtype: int64
- name: latency_ms
dtype: int64
- name: upstream_latency_ms
dtype: int64
- name: upstream_status
dtype: int64
- name: temperature
dtype: float64
- name: top_p
dtype: float64
- name: top_k
dtype: int64
- name: max_tokens
dtype: int64
- name: presence_penalty
dtype: float64
- name: repetition_penalty
dtype: float64
- name: error_type
dtype: string
- name: error_message
dtype: string
- name: ingest_source
dtype: string
splits:
- name: train
num_bytes: 80468220
num_examples: 726
download_size: 17821824
dataset_size: 80468220
configs:
- config_name: default
data_files:
- split: train
path: data/train-*