DeepFabric 4.8.3__py3-none-any.whl → 4.10.0__py3-none-any.whl

deepfabric/validation.py

@@ -1,6 +1,5 @@
 import time
 
-from .exceptions import ConfigurationError
 from .tui import get_tui
 
 
@@ -27,91 +26,39 @@ def calculate_expected_paths(mode: str, depth: int, degree: int) -> int:
 
 
 def validate_path_requirements(
-    mode: str,
-    depth: int,
-    degree: int,
-    num_steps: int,
-    batch_size: int,
-    loading_existing: bool = False,
+    mode: str,  # noqa: ARG001 - kept for API compatibility
+    depth: int,  # noqa: ARG001 - kept for API compatibility
+    degree: int,  # noqa: ARG001 - kept for API compatibility
+    num_samples: int | str,  # noqa: ARG001 - kept for API compatibility
+    batch_size: int,  # noqa: ARG001 - kept for API compatibility
+    loading_existing: bool = False,  # noqa: ARG001 - kept for API compatibility
 ) -> None:
     """
-    Validate that the topic generation parameters will produce enough paths.
+    Validate topic generation parameters (informational only, no longer errors).
+
+    When num_samples exceeds available paths, topics will cycle for even coverage.
 
     Args:
         mode: Generation mode ('tree' or 'graph')
         depth: Depth of the tree/graph
         degree: Branching factor
-        num_steps: Number of generation steps
-        batch_size: Batch size for generation
+        num_samples: Total samples to generate, or "auto"/percentage string
+        batch_size: Batch size for generation (not used in validation, kept for API compat)
         loading_existing: Whether loading existing topic model from file
-
-    Raises:
-        ConfigurationError: If validation fails
     """
-    if loading_existing:
-        # Can't validate existing files without loading them
-        return
-
-    expected_paths = calculate_expected_paths(mode, depth, degree)
-    required_samples = num_steps * batch_size
-
-    if required_samples > expected_paths:
-        # Alternative: provide exact combinations that use all paths
-        optimal_combinations = []
-        for test_steps in range(1, expected_paths + 1):
-            test_batch = expected_paths // test_steps
-            if test_steps * test_batch <= expected_paths and test_batch > 0:
-                optimal_combinations.append((test_steps, test_batch))
-
-        # Sort by preference (fewer steps first, then larger batches)
-        optimal_combinations.sort(key=lambda x: (x[0], -x[1]))
-
-        tui = get_tui()
-        tui.error(" Path validation failed - stopping before topic generation")
-
-        # Build recommendations - focus on optimal combinations rather than misleading individual params
-        recommendations = []
-
-        if optimal_combinations:
-            recommendations.append(
-                f"  • Use one of these combinations to utilize the {expected_paths} paths:"
-            )
-            for steps, batch in optimal_combinations[:3]:  # Show top 3
-                total_samples = steps * batch
-                recommendations.append(
-                    f"    --num-samples {steps} --batch-size {batch}  (generates {total_samples} samples)"
-                )
-
-        recommendations.extend(
-            [
-                f"  • Or increase --depth (currently {depth}) or --degree (currently {degree})",
-            ]
-        )
-
-        estimation_note = ""
-        if mode == "graph":
-            estimation_note = " (estimated - graphs vary due to cross-connections)"
-
-        error_msg = (
-            f"Insufficient expected paths for dataset generation:\n"
-            f"  • Expected {mode} paths: ~{expected_paths}{estimation_note} (depth={depth}, degree={degree})\n"
-            f"  • Requested samples: {required_samples} ({num_steps} steps × {batch_size} batch size)\n"
-            f"  • Shortfall: ~{required_samples - expected_paths} samples\n\n"
-            f"Recommendations:\n" + "\n".join(recommendations)
-        )
-
-        if mode == "graph":
-            error_msg += f"\n\nNote: Graph path counts are estimates. The actual graph may produce {expected_paths // 2}-{expected_paths * 2} paths due to cross-connections."
-
-        raise ConfigurationError(error_msg)
+    # No validation needed - generator handles all cases:
+    # - num_samples < paths: random subset
+    # - num_samples == paths: all paths used once
+    # - num_samples > paths: topics cycle for even coverage
+    pass
 
 
 def show_validation_success(
     mode: str,
     depth: int,
     degree: int,
-    num_steps: int,
-    batch_size: int,
+    num_samples: int | str,
+    batch_size: int,  # noqa: ARG001 - kept for API compatibility
     loading_existing: bool = False,
 ) -> None:
     """
@@ -121,21 +68,44 @@ def show_validation_success(
         mode: Generation mode ('tree' or 'graph')
         depth: Depth of the tree/graph
         degree: Branching factor
-        num_steps: Number of generation steps
-        batch_size: Batch size for generation
+        num_samples: Total samples to generate, or "auto"/percentage string
+        batch_size: Batch size for generation (not used in display, kept for API compat)
         loading_existing: Whether loading existing topic model from file
     """
     if loading_existing:
         return
 
     expected_paths = calculate_expected_paths(mode, depth, degree)
-    total_samples = num_steps * batch_size
-
     tui = get_tui()
+
+    # Handle dynamic num_samples (auto or percentage)
+    if isinstance(num_samples, str):
+        tui.success("Path Validation Passed")
+        tui.info(f"  Expected {mode} paths: ~{expected_paths} (depth={depth}, degree={degree})")
+        if num_samples == "auto":
+            tui.info(f"  Requested samples: auto (will use all ~{expected_paths} paths)")
+        else:
+            # Percentage string like "50%"
+            pct = float(num_samples[:-1])
+            estimated_samples = max(1, int(expected_paths * pct / 100))
+            tui.info(
+                f"  Requested samples: {num_samples} (~{estimated_samples} of {expected_paths} paths)"
+            )
+        if mode == "graph":
+            tui.info("  Note: Graph paths may vary due to cross-connections")
+        print()  # Extra space before topic generation
+        time.sleep(0.5)  # Brief pause to allow user to see the information
+        return
+
     tui.success("Path Validation Passed")
     tui.info(f"  Expected {mode} paths: ~{expected_paths} (depth={depth}, degree={degree})")
-    tui.info(f"  Requested samples: {total_samples} ({num_steps} steps x {batch_size} batch size)")
-    tui.info(f"  Path utilization: ~{min(100, (total_samples / expected_paths) * 100):.1f}%")
+    tui.info(f"  Requested samples: {num_samples}")
+
+    if num_samples > expected_paths:
+        cycles = (num_samples + expected_paths - 1) // expected_paths  # ceil division
+        tui.info(f"  Topic cycling: ~{cycles}x passes through topics")
+    else:
+        tui.info(f"  Path utilization: ~{(num_samples / expected_paths) * 100:.1f}%")
 
     if mode == "graph":
         tui.info("  Note: Graph paths may vary due to cross-connections")

{deepfabric-4.8.3.dist-info → deepfabric-4.10.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: DeepFabric
-Version: 4.8.3
+Version: 4.10.0
 Summary: Curate High Quality Datasets, Train, Evaluate and Ship
 Author-email: DeepFabric Team <oss@alwaysfurther.ai>
 License-File: LICENSE
@@ -11,7 +11,7 @@ Requires-Dist: componentize-py>=0.19.3
 Requires-Dist: datasets<5.0,>=3.0
 Requires-Dist: google-api-core>=2.0.0
 Requires-Dist: google-genai>=1.56.0
-Requires-Dist: huggingface-hub==0.36.0
+Requires-Dist: huggingface-hub>=1.3.1
 Requires-Dist: kagglehub>=0.3.0
 Requires-Dist: mermaid-py>=0.8.0
 Requires-Dist: ollama>=0.6.1
@@ -25,7 +25,7 @@ Requires-Dist: pyyaml>=6.0.1
 Requires-Dist: rich>=13.0.0
 Requires-Dist: sentencepiece>=0.1.99
 Requires-Dist: spin-sdk>=3.4.1
-Requires-Dist: transformers>=4.57.1
+Requires-Dist: transformers==5.0.0rc3
 Provides-Extra: dev
 Requires-Dist: bandit>=1.7.10; extra == 'dev'
 Requires-Dist: mermaid-py>=0.2.0; extra == 'dev'
@@ -172,12 +172,12 @@ generation:
     Provide context on when and why to use specific patterns or libraries.
     Ensure code is modular, testable, and maintainable.
 
+  # Agent mode is implicit when tools are configured
   conversation:
     type: cot      # basic | cot
     reasoning_style: agent      # freetext | agent (for cot)
-    agent_mode: single_turn     # single_turn | multi_turn (for agent)
-  
-  # Tool configuration (required for agent modes)
+
+  # Tool configuration (enables agent mode automatically)
   tools:
     spin_endpoint: "http://localhost:3000"  # Spin service for tool execution
     components:                 # Map component name to tool names
@@ -535,7 +535,6 @@ generation:
   conversation:
     type: cot
     reasoning_style: agent
-    agent_mode: single_turn
 
   tools:
     spin_endpoint: "http://localhost:3000"  # Spin service URL

{deepfabric-4.8.3.dist-info → deepfabric-4.10.0.dist-info}/RECORD

@@ -1,40 +1,40 @@
 deepfabric/__init__.py,sha256=DA4DT1nmF7VjQ00IA_8J_kSlqGYrK2EvP5H4XPzuKoA,1690
 deepfabric/__main__.py,sha256=Ojx6VFnAWx4wY52VehsWhW85IaEmPb8FP_sGvOk628g,113
 deepfabric/auth.py,sha256=7UEvk84XPx7wiXj0QfhIH_MJ5BlI7wQHodwgd4LPO80,12977
-deepfabric/builders.py,sha256=XKlKsAhsed2_M_uHft-VB-n8T1ZhAbIo_7XXc2mE3Ug,11503
-deepfabric/builders_agent.py,sha256=7xaXVNzmW_vBDkxJNKR_9HveljrdljwDAFx3iu-L_-M,48468
-deepfabric/cli.py,sha256=gf7HoXlLn-8SZHHmpNWI7dwXvv_VCzu7wCQraOJkKsc,51236
+deepfabric/builders.py,sha256=2-tJh0_yzGqcBn8qRy3ocd5qIhBciwLsGPiqP1PK75w,10821
+deepfabric/builders_agent.py,sha256=youDxc8tWJF0DbA3Al-6zcswx68gcI7z9Gz7hPsDTJo,28326
+deepfabric/cli.py,sha256=KGCR7AGEAbshfiFNu-MnNGdgqIVBF-vI2piHQzkRrv0,68750
 deepfabric/cloud_upload.py,sha256=WYaISQY1XxorNdL7_F_FYwQUPHGJr2Bb_bohAa5xpbY,27801
-deepfabric/config.py,sha256=5M8BhvWXGmC0rcvSlY3TKwGI8TIrYHL6xjQ95ITGs8o,34780
-deepfabric/config_manager.py,sha256=CIOJV121tBpH_V_ljwTenvyFO31yoohPSjW0yrHCD-w,9041
-deepfabric/constants.py,sha256=MwADziDmnt0zi9t9gG65EM7AJvIQP0FSsXgGj7Yqxm8,2578
+deepfabric/config.py,sha256=LJptRUykRGreirIwyUxF8rcverf1BD9hOPiiYNxkoEI,33990
+deepfabric/config_manager.py,sha256=c0N4Rb0P970khoiKgpN1ULDYV0bZEYIbec45cSfg92c,9334
+deepfabric/constants.py,sha256=nUc2Bf6Fu_c5biUy1nE5jCP9t6z08_ihWzarRTE1nwg,2827
 deepfabric/dataset.py,sha256=bZfx35A-dt0kMflgskU9Ge-NLVesq8xNKHsrxTnNn6Q,9740
-deepfabric/dataset_manager.py,sha256=fJ6VFG05FLTpmbkLKlnVTTi7aim8q7eWI1cgOmKaP5s,20461
+deepfabric/dataset_manager.py,sha256=hf0KQtYpLi4-P-a1czKNhqgrKsqFG8zJjfubbSBOuAo,24652
 deepfabric/error_codes.py,sha256=HGGWsahUTI8UG996C74X-XgNuaPX8RHo4gOidlaJql4,17630
 deepfabric/exceptions.py,sha256=pEg4YFQaDEWtBoJaSkxsJJoBBp2-6EE3M7m5H7R6i_8,1586
 deepfabric/factory.py,sha256=OCqo3w-eiYNWvK_I_egDZuWj192kf18yD3SPj8rrPxU,753
-deepfabric/generator.py,sha256=wdGxuKQOMGY8oEpa-YvXX2ceCnzRApDgzweLHgwtjlw,44226
-deepfabric/graph.py,sha256=JQ68GXnLymtR7ESfeZgdMh3YrReSPX5wEWEqXlkIR4Q,24175
+deepfabric/generator.py,sha256=sirx7qukTl_SM9RGYT7xPDP65L8BlNzZb7ykbvO4Sg8,68994
+deepfabric/graph.py,sha256=UehVpNnLKThNceteKUVfY401V160q7LFYm8cyTdE2bw,25787
 deepfabric/hf_hub.py,sha256=hw2CWqZ3CzyAzMo552VPZKVWtuv-j0TQ2_gV5K0AUto,7670
 deepfabric/kaggle_hub.py,sha256=CXVO1Lv3IRhdO0bp9_IQr6nUs-v5jOWi5k4EwPkbJmw,7927
 deepfabric/loader.py,sha256=YNTGZZE-POjR0BIlx6WCT4bIzf0T4lW_fQl7ev9UFqE,18584
 deepfabric/metrics.py,sha256=txqmXDM_r6cWPjdnnEjoA5xJkCHxFrjKWTpihE_jimA,6129
 deepfabric/progress.py,sha256=3XQQrf2pUZlyd-8eRcNATH1v0Oi8JMedVHGbhPcca-8,9354
-deepfabric/prompts.py,sha256=XKFaoiT9G_t7z3VhWNr1xsWx78I-2kzq6wErc6DW1eI,15397
-deepfabric/schemas.py,sha256=N1cTvXuAyV8r8YS5DSAcFgpfxF0AqVGJbbOpeT5H72g,37881
+deepfabric/prompts.py,sha256=pAFkw_0WAcYdzjqkHgywyLiSEdDlRdZwmyykUC307Gk,13872
+deepfabric/schemas.py,sha256=ebSwaXCu3lwshFGEhOhuAAUQA3deaYWXUcoRaeslqLc,37880
 deepfabric/stream_simulator.py,sha256=GzvAxWxHVsuTwgXlqwXNfrTUDn6sND2kJOoQuYg88FA,3028
 deepfabric/topic_manager.py,sha256=6YxMO6dQHaGyxghsI8iNJGP1miaekBe5Mh1WdYeLqdI,11164
-deepfabric/topic_model.py,sha256=i_wYpw2kUl8NLodOSaqNu-C4_d6caYT1kPe_vkKjoyw,707
-deepfabric/tree.py,sha256=Kxl2iLHU55xPq2MwdoLM0-M2nZRx51bRj9FM36jqs-M,14933
-deepfabric/tui.py,sha256=9ETtGFQk26U9PQ2b5foplVYDKxaFGd-8UqK7uSKyHwE,50480
+deepfabric/topic_model.py,sha256=3RupTQnEYfKSLASV79tfEhBXBrm5VT3nUcpl2EyK7Hk,1654
+deepfabric/tree.py,sha256=0J5KuHJrVEfF4F6zM9o2-ppAEzolay_9fDloZCHDN6c,15791
+deepfabric/tui.py,sha256=hUYK00czktdPT5W3GJQ5svObbDYMAAnou3u-YDOEC-o,52608
 deepfabric/update_checker.py,sha256=AUa9iUdkGNzu7tWkQRxIlF19YRmKLetwxu-Ys2ONS8Y,5145
-deepfabric/utils.py,sha256=a9G6VTw52UdddTFoMw-JjunjawtPN54N275-XGPL2cQ,5822
-deepfabric/validation.py,sha256=1x1X_45kyI0w_FCdUiNdvy4LQu3B0KVR-fyvLkrKEGw,5125
+deepfabric/utils.py,sha256=nVDNQnNQlfua8mMpKC7gk9EB3bYXbNUk6Sq39k1mN7s,12507
+deepfabric/validation.py,sha256=D6H_L_0mDtiI0s6DLXxkR1BickOw7HGM2al8u2xhKZk,4355
 deepfabric/evaluation/__init__.py,sha256=7xMLmYXaNC1U7qf88S9fMxWTABoDRiOcimSYfCt_PSo,1224
 deepfabric/evaluation/evaluator.py,sha256=qNowle5v2ukDJ11igNOCParlBfXT8QUeOvXx6sSJ_Ug,34480
 deepfabric/evaluation/inference.py,sha256=y7JA0IsBDwe0sJzVQeItYHAV5wUJn6Bjp1Wsp3r7qYQ,7644
 deepfabric/evaluation/metrics.py,sha256=ITNevYj7CBXzYs-rYhsihO6-rE9n30CYRaVUfdTbcFQ,12026
-deepfabric/evaluation/parser.py,sha256=AXyiCtNV4rueZQxLE_GqqkFNeDAewGoC--0vXHW-jW8,10603
+deepfabric/evaluation/parser.py,sha256=KLRP5cI6y2wqtbX2kPnyEJiYHqMk5qDa1toqJrKetb4,10604
 deepfabric/evaluation/backends/__init__.py,sha256=GqC0FfpWmtgJmjHd0kVKNg7g-NjhRoh5h2MtAoOhUOY,827
 deepfabric/evaluation/backends/llm_eval_backend.py,sha256=4jp5tnTp7v_0pHCGhcPbI55ig79-eVxdzooesi2PymA,18827
 deepfabric/evaluation/backends/ollama_backend.py,sha256=mtPp1JtIDRjb76X_rTa1jS1ETzMjte8t3WJjuYV1oDQ,4372
@@ -69,8 +69,8 @@ deepfabric/training/api_key_prompt.py,sha256=pSIMX3eDGyV9x_r7MHE4TyIsIB2SqYb8gKC
 deepfabric/training/callback.py,sha256=5zdifbHA2PWILHl2cVFyO65aW7cGAQhcvDqm3s8_I0Q,13221
 deepfabric/training/dataset_utils.py,sha256=klx8DoawEwuMigBDP-RpMAfe7FvYxRbhj599MErxBr4,7313
 deepfabric/training/metrics_sender.py,sha256=ZCyvMv5hRu8XJnQYVGXJ9wh7HEMJ0l3Ktyi8_etOpZs,10833
-deepfabric-4.8.3.dist-info/METADATA,sha256=ppBY0UdQd2bybvZF0HcXivMwFnHGEn_Nk6kCsDFBR6c,20536
-deepfabric-4.8.3.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-deepfabric-4.8.3.dist-info/entry_points.txt,sha256=zatevils13hfs8x29_vmUyivQ6rTtq7hE2RBusZw1Fo,50
-deepfabric-4.8.3.dist-info/licenses/LICENSE,sha256=-qRt8wmrhQ9aMf7KhmZXc2vrTETYZF-6_T1KCeUhvHY,11340
-deepfabric-4.8.3.dist-info/RECORD,,
+deepfabric-4.10.0.dist-info/METADATA,sha256=bybLquu722z-9xb40wCLHe6QIliwAYSW7oHONs_D02Y,20498
+deepfabric-4.10.0.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+deepfabric-4.10.0.dist-info/entry_points.txt,sha256=zatevils13hfs8x29_vmUyivQ6rTtq7hE2RBusZw1Fo,50
+deepfabric-4.10.0.dist-info/licenses/LICENSE,sha256=-qRt8wmrhQ9aMf7KhmZXc2vrTETYZF-6_T1KCeUhvHY,11340
+deepfabric-4.10.0.dist-info/RECORD,,