Controlled Psycholinguistic Experiments with Large Language Models

---

Documentation for package ‘PsyLingLLM’ version 0.4.0

• DESCRIPTION file.
• Package NEWS.

Help Pages

adaptive_feedback_experiment	Run an Adaptive Feedback Experiment
build_body_pass2_structural	Build Pass-2 body
build_registry_entry_from_analysis	Build a registry entry from analysis results
build_standardized_input	Build standardized Pass-2 headers & body from Pass-1 raw inputs
cat_slowly	Print lines gradually with delay (console-safe)
classify_generation_interface	Classify LLM generation interface from URL (robust and version-agnostic)
common_error_blacklist	Common provider error-field blacklist (path segments only)
compare_path_line	Compare two optional scalar strings (paths). Return a structured line and a boolean flag.
complete_headers_for_pass2	Complete headers minimally for Pass-2 (keeps user values)
conversation_experiment	Run a Multi-turn Conversation Experiment
conversation_experiment_with_feedback	Run an Adaptive Feedback Experiment
default_keyword_lexicon	Default keyword lexicon for ns/st scoring
default_optional_keys	Default optional keys for common LLM providers (explicit removal first)
detect_embedded_think_tag	Detect embedded <think> tags in model output
ensure_registry_header	Ensure registry directory exists and header is present
explain_alias	Explain how an alias was derived from a model id
extract_pass2_templates	Tolerant extraction of Pass-2 templates (headers and body)
extract_process_metrics	Extract Standard Process Metrics from RAW Trial Results
extract_token_probabilities	Extract Token Probabilities from RAW Trial Results
extract_usage_fields	Extract usage field names from a parsed non-streaming response
factorial_trial_experiment	Run Factorial Trial Experiment (registry-driven, assistant-native)
find_placeholder_path	Recursively find the first path to a placeholder string
flatten_json_paths	Flatten a JSON-like object into (path, key, value) rows
format_registration_preview	Render a registry entry preview (verbatim from the registry node)
generate_llm_experiment_list	Generate LLM Experiment List (CSV→XLSX conversion, UTF-8 safe)
generate_llm_factorial_experiment_list	Generate LLM Experiment List for Factorial Design (UTF-8 safe)
get_model_config	Resolve a model name (id or alias) to a registry entry (new registry-aware)
get_registry_entry	Retrieve a registry interface entry (inputs / outputs / streaming)
get_registry_path	Get user registry path (cross-platform, unified)
get_trial_raw	Inspect One Archived RAW Trial Record
has_nonempty_text	Does a value contain any non-empty textual content?
infer_role_mapping_from_body	Infer role mapping from a chat-style LLM request body
infer_structure_from_placeholder	Infer structure from where '"\${CONTENT}"' lives
inject_optional_params	Inject optional params into a body using a whitelist
is_invalid_candidate	Hard filter: blacklist, nullish, booleans, numeric-only, too-short Segment-exact blacklist (compares path segments, not substrings).
is_key	Utility: check if a path segment is a named key
is_message_object	Check if an object is a message object
list_get_by_path	Get a nested value by a mixed path (character keys and integer indices)
llm_caller	Call an LLM via the registry
llm_register	Register an LLM Endpoint with Pass-2 Validation
load_registry	Load the system LLM registry
make_alias	Generate a compact alias for a model id
make_pass2_probe_inputs	Build Pass-2 EFFECTIVE probe inputs (no forced system; structural role inference)
merge_defaults_for_probe	Merge default generation parameters into a request body (probe only)
multi_model_experiment	Run Trials Across Multiple Models (registry-driven, assistant-native)
normalize_history_messages	Normalize conversation history into message objects
normalize_path_key_with_regex	Normalize any path spec to a flat key AND, if needed, a regex with wildcard index
normalize_provider_label	Normalize a free-form provider label
normalize_type_label	Normalize a free-form type string (strip a leading '@' if present)
probe_llm_streaming	Probe an LLM endpoint (non-stream + streaming, SSE-aware)
PsyLingLLM_Schema	PsyLingLLM Schema (v1)
register_endpoint_offline	Register an LLM endpoint offline (supports reasoning / thinking path)
register_endpoint_to_user_registry	Upsert endpoint entry into the user registry (safe merge)
render_pass2_path_consistency_report	Render a structured "[Pass-2 Verified Ports]" report section.
save_experiment_results	Save Experiment Results to File
score_candidates_ns	Rank non-streaming JSON paths for answer/think (segment-aware, conflict-guarded)
score_candidates_st	Rank streaming (SSE) candidates for answer/think (temporal aggregation, delta-aware, conflict-guarded)
stream_reconstruct_text	Reconstruct stream text from raw_json using exact or wildcard-index matching Falls back to regex (for ".." wildcard) if exact match finds nothing.
strip_optionals_from_body	Strip optional runtime parameters from a request body
sub_placeholders	Substitute placeholders '"\${API_KEY}"', '"\${CONTENT}"', '"\${ROLE}"' robustly
trial_experiment	Run LLM Trial Experiment
update_progress_bar	Update progress bar for an experiment
validate_experiment_config	Validate experiment configuration and resolve API URL
validate_registry_entry	Validate structure of a registry entry
validate_registry_schema	Validate a PsyLingLLM Registry Document