Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
242 commits
Select commit Hold shift + click to select a range
40047bd
[Feature] Add easi library - Phase 1 implementation
oscarqjh Feb 11, 2026
cdbf211
[Feature] Add easi library - Phase 2 implementation
oscarqjh Feb 12, 2026
091d38c
[Refactor] Separate generic AI2-THOR bridge from EB-Alfred task logic
oscarqjh Feb 12, 2026
1ad66a0
[Fix] Updated dependencies for ebalfred
oscarqjh Feb 13, 2026
52d2f30
[Feat] Added EBAlfred, working on dummy agent
oscarqjh Feb 14, 2026
347a0ca
[Feat] Added auto env install
oscarqjh Feb 14, 2026
65ec22b
[Feat] Add LLM response schemas with BaseResponseSchema
oscarqjh Feb 14, 2026
9c8dd29
[Feat] Add LLM kwargs parser, backend validation, model string builder
oscarqjh Feb 14, 2026
e1a85ba
[Feat] Add LLMClient wrapping LiteLLM + Instructor
oscarqjh Feb 14, 2026
ab27e7e
[Feat] Add ServerManager for vLLM subprocess lifecycle
oscarqjh Feb 14, 2026
3e7010c
[Feat] Add --backend, --model, --port, --llm-kwargs CLI args
oscarqjh Feb 14, 2026
7856803
[Feat] Integrate LLMClient + ServerManager into EvaluationRunner
oscarqjh Feb 14, 2026
7231f0b
[Feat] Add [llm] optional dependency group (litellm + instructor)
oscarqjh Feb 14, 2026
e7f73c1
[Test] Add import smoke tests and full LLM pipeline integration test
oscarqjh Feb 14, 2026
cb972eb
[Fix] Address code review findings for LLM inference pipeline
oscarqjh Feb 14, 2026
c57b054
[Refactor] Introduce AgentMemory architecture + chat_history=True sup…
oscarqjh Feb 14, 2026
ad58088
[Refactor] Add response_format pass-through for structured LLM output
oscarqjh Feb 15, 2026
33aa620
[Fix] Upated task yaml for ebalfred
oscarqjh Feb 15, 2026
ed77434
[Refactor] Move dynamic_action_space handling from runner to task hook
oscarqjh Feb 15, 2026
4483419
[Feat] Enrich run outputs with LLM responses, instruction, and full c…
oscarqjh Feb 15, 2026
c8648b3
[Feat] Add retry and resume support for evaluation runs
oscarqjh Feb 15, 2026
912cfb0
[Fix] Default --agent to react instead of dummy
oscarqjh Feb 15, 2026
36aff7b
[Feat] Add per-episode retry with simulator re-launch and fix max_ste…
oscarqjh Feb 15, 2026
02ed1e2
[Feat] Add EB-Navigation benchmark integration
oscarqjh Feb 15, 2026
9c8d7ee
[Refactor] Replace YAML action_space with BaseTask._build_action_space()
oscarqjh Feb 15, 2026
19f8e8e
[Feat] Implement real AI2-THOR v5.0.0 bridge and switch to Linux64 pl…
oscarqjh Feb 15, 2026
ca6cfff
[Fix] Align EB-Navigation prompts with EmbodiedBench source exactly
oscarqjh Feb 15, 2026
c7c2d80
[Fix] Updated ebnavigation prompt
oscarqjh Feb 16, 2026
d73cb20
[Feat] Add EB-Habitat benchmark integration
oscarqjh Feb 16, 2026
b9d339c
[Feat] Add --redownload flag for dataset re-download
oscarqjh Feb 16, 2026
8a46da3
[Feat] Add retry with exponential backoff for HuggingFace dataset dow…
oscarqjh Feb 16, 2026
d543878
[Fix] Fix habitat-sim conda env spec to use separate feature packages
oscarqjh Feb 16, 2026
1128c6e
[Fix] Use Python 3.9 for habitat-sim 0.3.0 (no 3.10 builds available)
oscarqjh Feb 16, 2026
751ceea
[Fix] Only validate habitat_sim import for simulator env
oscarqjh Feb 16, 2026
f43079b
[Fix] Restore side-effect imports, trim unused imports, and add missi…
oscarqjh Feb 16, 2026
907879b
[Fix] Fix EB-Habitat runtime deps and rendering config
oscarqjh Feb 16, 2026
5a838e8
[Fix] Pass HF dataset path to EB-Habitat bridge via reset_config
oscarqjh Feb 16, 2026
32e483f
[Refactor] Make EB-Habitat env constants configurable via task YAML
oscarqjh Feb 16, 2026
60eddb2
[Refactor] Make EB-Navigation env constants configurable via task YAML
oscarqjh Feb 16, 2026
d6a63a9
[Refactor] Make EB-Alfred env constants configurable via task YAML
oscarqjh Feb 16, 2026
ffbed42
[Feat] Add configurable bounding box drawing for EB-Navigation observ…
oscarqjh Feb 16, 2026
321de2a
[Feat] Add configurable bounding box drawing for EB-Alfred observations
oscarqjh Feb 16, 2026
d55ab73
[Chore] Add json file
oscarqjh Feb 16, 2026
e948a7a
[Feat] Make EB-Habitat gpu_device_id configurable via task YAML
oscarqjh Feb 16, 2026
8f0d947
[Fix] Add generation_kwargs to YAML configs and fix empty plan handling
oscarqjh Feb 17, 2026
9f33a63
[Fix] Drop unsupported LLM params per provider and add run config tra…
oscarqjh Feb 17, 2026
54788b8
[Fix] Disable Habitat episode shuffle and add model name to run direc…
oscarqjh Feb 17, 2026
d48cf21
[Feat] Add CoppeliaSim V4.1.0 simulator integration
oscarqjh Feb 18, 2026
5c9e80c
[Fix] Trust DISPLAY env var for real screen detection
oscarqjh Feb 18, 2026
30dc5c7
[Feat] Add EB-Manipulation benchmark integration with configurable ta…
oscarqjh Feb 18, 2026
ccdc800
[Feat] Add YAML template inheritance utilities (deep_merge + resolve_…
oscarqjh Feb 18, 2026
bff7d0e
[Feat] Integrate YAML extends into registry and BaseTask config loading
oscarqjh Feb 18, 2026
7ee5a47
[Refactor] Migrate EB-Manipulation YAMLs to extends template inheritance
oscarqjh Feb 18, 2026
80f4fbb
[Refactor] Migrate EB-Navigation YAMLs to extends template inheritance
oscarqjh Feb 18, 2026
5438ff9
[Refactor] Migrate EB-Alfred YAMLs to extends template inheritance
oscarqjh Feb 18, 2026
aa0d783
[Refactor] Migrate EB-Habitat YAMLs to extends template inheritance
oscarqjh Feb 18, 2026
4ddd6c9
[Refactor] Auto-capture config.json from __init__ args via inspect
oscarqjh Feb 19, 2026
2bce3b2
[Feat] Rename easi run → easi start with multi-task support
oscarqjh Feb 19, 2026
e20305a
[Feat] Stream vLLM server output through TRACE logger instead of log …
oscarqjh Feb 19, 2026
1751b21
[Chore] Updated pyproject.toml
oscarqjh Feb 19, 2026
22bfe0b
[Fix] Route trust_remote_code and other vLLM flags to server kwargs
oscarqjh Feb 19, 2026
4fad44b
[Fix] Fixed vllm and transformers version
oscarqjh Feb 19, 2026
605f4e8
[Feat] Add parallel evaluation runner with --num-parallel flag
oscarqjh Feb 19, 2026
d522a10
[Fix] Unify resume logic for sequential and parallel runners
oscarqjh Feb 19, 2026
f4edcf6
[Feat] Add TDW v1.11.23 simulator integration for HAZARD benchmark
oscarqjh Feb 20, 2026
b36d99c
feat: add EpisodeRecord and pluggable aggregate_results to BaseTask
oscarqjh Feb 20, 2026
789d022
feat: integrate aggregate_results into runners, nest metrics in summa…
oscarqjh Feb 20, 2026
5413684
[Chore] Remove redundant comments in prompt_builder and react_agent
oscarqjh Feb 20, 2026
5835fae
[Docs] Update CLAUDE.md and add AGENT.md for AI-assisted development
oscarqjh Feb 20, 2026
8952492
[Fix] Vendor amsolver, tools, vlm into EB-Manipulation to remove exte…
oscarqjh Feb 20, 2026
cf16fad
Merge branch 'dev' into worktree-vendor-hazard
oscarqjh Feb 20, 2026
3c7a8ab
[Feat] Vendor HAZARD source and add HAZARD task integration
oscarqjh Feb 20, 2026
4690439
[Feat] Auto-download and auto-launch TDW build for HAZARD
oscarqjh Feb 20, 2026
36c3486
[Feat] Auto-download TDW build, auto-launch in bridge, and misc impro…
oscarqjh Feb 20, 2026
2b8be6b
[Feat] Auto-download TDW build, auto-launch in bridge, and misc impro…
oscarqjh Feb 21, 2026
4f3a176
[Docs] Add detailed CLI reference
oscarqjh Feb 21, 2026
c31ad61
[Feat] Add OmniGibson simulator integration for BEHAVIOR-1K
oscarqjh Feb 21, 2026
ee9eadf
[Fix] NFS workaround for OmniGibson: copy Python to /tmp to fix /proc…
oscarqjh Feb 22, 2026
e94db3a
feat: add RenderPlatform abstraction with auto/native/xvfb/egl/headless
oscarqjh Feb 23, 2026
6e2dad6
feat: add render platform properties to BaseEnvironmentManager
oscarqjh Feb 23, 2026
904b110
feat: SubprocessRunner uses RenderPlatform for command wrapping
oscarqjh Feb 23, 2026
463fd84
feat: declare render platform preferences in all simulator env managers
oscarqjh Feb 23, 2026
0e9df38
feat: add --render-platform to CLI with validation and YAML support
oscarqjh Feb 23, 2026
46582be
test: add render platform integration and config persistence tests
oscarqjh Feb 23, 2026
715142a
chore: remove redundant md files
oscarqjh Feb 23, 2026
b52c56e
refactor: replace raw env var dicts with EnvVars dataclass
oscarqjh Feb 23, 2026
e8395ba
feat: support custom render platform classes via simulator manifest
oscarqjh Feb 23, 2026
2c21170
feat: move CoppeliaSim platform-specific env vars to custom render pl…
oscarqjh Feb 23, 2026
1b6c8a3
[Feat] Add option to add object state history to prompt
oscarqjh Feb 23, 2026
3c33f6e
chore: add yolo11n.pt to gitignore
oscarqjh Feb 23, 2026
d97a815
fix: always set Qt plugin path and control CoppeliaSim headless via p…
oscarqjh Feb 23, 2026
db88011
fix: commit files
oscarqjh Feb 23, 2026
e98c381
chore: remove redundant comments
oscarqjh Feb 23, 2026
da099ae
fix: removed redundant code
oscarqjh Feb 23, 2026
abf0ac5
chore: updated cli reference
oscarqjh Feb 23, 2026
004d93e
feat: add OmniGibson render platform classes (native + auto)
oscarqjh Feb 23, 2026
942a99c
feat: vendor minimal rearrangement environment code
oscarqjh Feb 24, 2026
f4ba251
feat: add rearrangement 84-action space module
oscarqjh Feb 24, 2026
928c595
feat: add rearrangement task YAMLs for 5 splits
oscarqjh Feb 24, 2026
9bdb41e
feat: add AI2THORRearrangement2023Task with 6 paper metrics
oscarqjh Feb 24, 2026
4f3fdab
feat: add rearrangement bridge and prompt builder
oscarqjh Feb 24, 2026
c6a78c8
test: add prompt builder and bridge path tests
oscarqjh Feb 24, 2026
1d2cf4a
fix: use correct argparse dest name in e2e test Namespaces
oscarqjh Feb 24, 2026
17494ce
fix: remove xvfb system deps
oscarqjh Feb 24, 2026
08123bd
fix: OmniGibson simulator now passes smoke test
oscarqjh Feb 24, 2026
63a3b7f
Merge remote-tracking branch 'origin/dev' into dev
oscarqjh Feb 24, 2026
59a2aa7
fix: replace torch.lazy_property with functools.cached_property
oscarqjh Feb 24, 2026
2dcfd64
fix: add allenact as additional_deps, revert inlined stubs
oscarqjh Feb 24, 2026
cd35ade
fix: add lru-dict dep and fix bridge reset
oscarqjh Feb 24, 2026
6e1ff8b
fix: AI2-THOR rearrangement prompt issues and v5 compatibility
oscarqjh Feb 24, 2026
e954703
feat: dynamic sensor descriptions in prompt and depth image capture
oscarqjh Feb 24, 2026
b4127ab
chore: remove obsolete backward-compat tests
oscarqjh Feb 24, 2026
33da190
fix: aggregate_results crashes on non-numeric episode_results values
oscarqjh Feb 24, 2026
f3997bb
fix: aggregate only official metrics + runtime stats in summary
oscarqjh Feb 24, 2026
1840cf2
fix: warn agent that all objects are evaluated, not just listed ones
oscarqjh Feb 24, 2026
1d18edd
fix: remove stale object-list reference from prompt warning
oscarqjh Feb 24, 2026
d214f7f
feat: add AI2-THOR v3.3.5 simulator (arm mode) for ManipulaTHOR
oscarqjh Feb 25, 2026
0490a04
feat: add ManipulaTHOR Arm Point Navigation task integration
oscarqjh Feb 26, 2026
100467b
feat: add runtime and data_dir fields to SimulatorEntry
oscarqjh Mar 2, 2026
5832cde
feat: add DockerEnvironmentManager base class
oscarqjh Mar 2, 2026
6fcb527
feat: add Docker launch mode to SubprocessRunner
oscarqjh Mar 2, 2026
f5fe1e0
feat: add docker system dependency check
oscarqjh Mar 2, 2026
6cb48b0
feat: add Matterport3D simulator scaffold (first Docker user)
oscarqjh Mar 2, 2026
1216467
feat: wire Docker runtime detection into EvaluationRunner
oscarqjh Mar 2, 2026
874e100
feat: CLI support for Docker env install/check/list and sim test
oscarqjh Mar 3, 2026
63a6cab
feat(lhpr_vln): vendor SceneSimulator, config, and metrics from LH-VLN
oscarqjh Mar 3, 2026
d87ae9c
fix(lhpr_vln): restore metrics docstrings, remove unused imports
oscarqjh Mar 3, 2026
3980763
feat(lhpr_vln): add action space constants
oscarqjh Mar 3, 2026
4da8634
feat(lhpr_vln): add bridge wrapping vendored SceneSimulator
oscarqjh Mar 3, 2026
2463e68
feat(lhpr_vln): add task class with all 8 metrics + contest score
oscarqjh Mar 3, 2026
2d8c909
feat(lhpr-vln): add YAML task configs for val and test splits
oscarqjh Mar 3, 2026
9f1c38a
feat(habitat-sim): widen version spec to support 0.3.0 and 0.3.1
oscarqjh Mar 3, 2026
e4af2a4
feat(lhpr-vln): add prompt builder for LLM-based VLN agent
oscarqjh Mar 3, 2026
3c0294b
test(lhpr-vln): add unit tests for task, metrics, prompts, and actions
oscarqjh Mar 3, 2026
23fece5
feat(lhpr-vln): align prompt builder with EmbodiedBench format + togg…
oscarqjh Mar 3, 2026
a483a61
fix: add hf_data_dir support to prevent HM3D scene files from polluti…
oscarqjh Mar 3, 2026
a85ed84
fix(lhpr-vln): add future annotations for Python 3.9 compatibility
oscarqjh Mar 3, 2026
4665fa9
fix(lhpr-vln): use software rendering by default (gpu_device_id: -1)
oscarqjh Mar 3, 2026
0c32911
fix(lhpr-vln): clarify that stop advances subtask, not ends episode
oscarqjh Mar 3, 2026
eec28bd
feat: add configurable action_history_len to LHPR-VLN prompt builder
oscarqjh Mar 4, 2026
f6ae016
fix(lhpr-vln): improved prompt
oscarqjh Mar 4, 2026
83a7211
fix(lhpr-vln): updated action history len
oscarqjh Mar 4, 2026
369f1a6
feat: add robot-type grouping and action_history_len to LHPR-VLN
oscarqjh Mar 4, 2026
2b3168f
feat: add cuda_visible_devices to ServerManager
oscarqjh Mar 4, 2026
c2c83f9
feat: add default vLLM flags (prefix caching, quiet logs)
oscarqjh Mar 4, 2026
cccf861
feat: add trace logging to ServerManager for debugging
oscarqjh Mar 4, 2026
be385c2
feat: add MultiServerManager for multi-instance vLLM
oscarqjh Mar 4, 2026
b6d414b
feat: add --vllm-instances, --vllm-gpus, --sim-gpus CLI args
oscarqjh Mar 4, 2026
49c4943
feat: pass vllm_instances, vllm_gpus, sim_gpus through runners
oscarqjh Mar 4, 2026
364d922
feat: enable parallel vLLM evaluation with multi-instance support
oscarqjh Mar 4, 2026
f2610f9
feat: add simulator GPU isolation via sim_gpus
oscarqjh Mar 4, 2026
c58a757
feat: add GPU allocation validation for parallel vLLM
oscarqjh Mar 4, 2026
b0427b0
fix: update tests for new GPU allocation kwargs
oscarqjh Mar 4, 2026
33b0d93
fix: add port auto-probing and partial-failure cleanup to MultiServer…
oscarqjh Mar 4, 2026
cd84c96
fix: improve parallel vLLM robustness and observability
oscarqjh Mar 5, 2026
ddc27a1
fix: add warning
oscarqjh Mar 5, 2026
a49b9f9
docs: updated cli-reference
oscarqjh Mar 5, 2026
dab2eec
Merge branch 'EvolvingLMMs-Lab:main' into dev
oscarqjh Mar 9, 2026
ebb591f
feat: add habitat-sim v0.1.7 simulator integration (Python 3.7)
oscarqjh Mar 7, 2026
e2f7634
fix: use Python 3.8 for habitat-sim v0.1.7 (uv requires >=3.8)
oscarqjh Mar 7, 2026
aa13165
feat: add VLN-CE R2R task integration
oscarqjh Mar 7, 2026
5da4f0d
fix: resolve VLN-CE R2R integration issues found during testing
oscarqjh Mar 8, 2026
abb36e0
fix: use correct AgentMemory attribute names in VLN-CE prompt builder
oscarqjh Mar 8, 2026
cffec62
fix: correct AgentMemory API usage in VLN-CE prompt builder
oscarqjh Mar 8, 2026
14e1bbe
feat: add VLN-CE RxR task integration
oscarqjh Mar 8, 2026
3310eec
docs: add EASI standard prompt format reference
oscarqjh Mar 8, 2026
bc6021c
fix: align all in-scope prompt builders with EASI standard format
oscarqjh Mar 8, 2026
74a5303
fix(vlnce): avoid duplicate distance-to-goal in environment feedback
oscarqjh Mar 8, 2026
825f8eb
fix: resolve 4 bugs found in code review
oscarqjh Mar 9, 2026
5043a0d
refactor: rename --vllm-gpus/instances to --llm-gpus/instances
oscarqjh Mar 9, 2026
164a746
feat: add BaseModelServer ABC and message helpers for custom models
oscarqjh Mar 9, 2026
6276302
feat: add custom model registry with YAML manifest auto-discovery
oscarqjh Mar 9, 2026
c4c3602
feat: wire custom backend into ServerManager and LLM utils
oscarqjh Mar 9, 2026
74c8833
feat: add FastAPI HTTP server wrapper for custom models
oscarqjh Mar 9, 2026
05833ba
feat: support custom backend in evaluation runners
oscarqjh Mar 9, 2026
4efab38
feat: add 'easi model list/info' CLI subcommands
oscarqjh Mar 9, 2026
029103e
feat: add echo model as built-in example for custom server pipeline
oscarqjh Mar 9, 2026
372b4be
feat: add Qwen3-VL custom model example with message format conversion
oscarqjh Mar 9, 2026
77667dd
fix: pass actual backend to ServerManager instead of hardcoded 'vllm'
oscarqjh Mar 9, 2026
99556ad
fix: multiple bugs in custom model server pipeline
oscarqjh Mar 9, 2026
c3876bc
feat: add per-worker labels to simulator subprocess logs
oscarqjh Mar 9, 2026
7587820
feat: add sticky progress bar for evaluation runs
oscarqjh Mar 9, 2026
ce5ba7f
fix: increase default LLM timeout for local backends to 600s
oscarqjh Mar 9, 2026
3ce43a7
fix: include model_path in run directory name for custom backend
oscarqjh Mar 9, 2026
5716c39
docs: update CLI reference with custom model backend details
oscarqjh Mar 9, 2026
d0bdf39
fix: address code review issues in custom inference server
oscarqjh Mar 9, 2026
ab04b0b
fix: audit issues 1-2 + add easi ps command
oscarqjh Mar 10, 2026
e6fedfa
fix: process cleanup, worker safety, aggregation guard, resume data
oscarqjh Mar 10, 2026
6f95059
fix: handle simulator restart failure in retry path
oscarqjh Mar 10, 2026
972e46e
feat: add XorgPlatform render platform
oscarqjh Mar 10, 2026
0bf9c84
feat: add XorgManager for Xorg server lifecycle
oscarqjh Mar 10, 2026
5e9ffe3
fix: use sudo kill for sudo-launched Xorg processes in stop()
oscarqjh Mar 10, 2026
c6486b1
feat: add per-worker GPU pinning via round-robin
oscarqjh Mar 10, 2026
243eac2
feat: integrate XorgManager lifecycle into evaluation runners
oscarqjh Mar 10, 2026
7c1068a
docs: add xorg to render platform CLI options and reference
oscarqjh Mar 10, 2026
a99a810
feat: register xorg as supported render platform for X11 simulators
oscarqjh Mar 10, 2026
1dba15a
refactor: add lifecycle hooks to RenderPlatform, XorgPlatform owns Xo…
oscarqjh Mar 10, 2026
567298d
fix: asyncio deprecation, logger convention, litellm import scope
oscarqjh Mar 10, 2026
2b7011a
refactor: extract render_platform module into render_platforms package
oscarqjh Mar 11, 2026
a2acb53
fix: initialize _render_platform in ParallelRunner
oscarqjh Mar 11, 2026
537c5df
feat: add CoppeliaSimXorgPlatform for GPU-accelerated rendering
oscarqjh Mar 11, 2026
869b009
feat: add xorg render platform support to sim test CLI
oscarqjh Mar 11, 2026
6433a30
refactor: migrate render strategy integration to bindings and adapters
oscarqjh Mar 11, 2026
095fcff
fix: make sim test use binding-aware render launch
oscarqjh Mar 11, 2026
3450c34
feat: log resolved render platform for sim test and start commands
oscarqjh Mar 12, 2026
a71cb0f
refactor: add log_name property to RenderPlatform to deduplicate logging
oscarqjh Mar 12, 2026
7a1344c
fix: ensure render platform setup/teardown lifecycle in all code paths
oscarqjh Mar 12, 2026
3d9e435
feat: add REVERIE-CE task (navigation-only)
oscarqjh Mar 12, 2026
811d5f9
feat: add trajectory video renderer for post-eval analysis
oscarqjh Mar 12, 2026
97cc3ef
feat: add _get_topdown_map and _get_episode_meta hooks to BaseBridge
oscarqjh Mar 13, 2026
8629bf8
feat: add topdown map and episode meta hooks to VLNCEBridge
oscarqjh Mar 13, 2026
8d13aa6
feat: add easi analyze trajectory CLI command
oscarqjh Mar 13, 2026
df1af98
feat: add topdown map, episode meta, and agent position to EBHabitatB…
oscarqjh Mar 13, 2026
96bc896
fix: correct sim access chain in EBHabitatBridge (was missing one .en…
oscarqjh Mar 13, 2026
c9268d7
feat: use progress bar for trajectory video generation
oscarqjh Mar 14, 2026
fc6ed05
feat: replace --max-episodes with --episodes filter (IDs, ranges, mixed)
oscarqjh Mar 15, 2026
465d35d
feat: vLLM v0.17+ compat and InternVL3 custom model server
oscarqjh Mar 18, 2026
8a7679e
chore: add internvl3 template
oscarqjh Mar 18, 2026
f4fcce0
fix: dont explicitly set egl vendor path
oscarqjh Mar 25, 2026
3c53bbe
feat: added unseen setting to lhvln
oscarqjh Apr 7, 2026
3fd1144
fix: fix gpu assignment for habitat sim
oscarqjh Apr 7, 2026
cdfa7a1
feat: added more sim config for lh-vln
oscarqjh Apr 7, 2026
53ad1cb
fix: prompt bug fixed for lhvln
oscarqjh Apr 8, 2026
7f7792c
fix: prompt bug
oscarqjh Apr 8, 2026
de6e029
feat: add new prompt format to lhvln
oscarqjh Apr 9, 2026
c932b69
feat: consecutive fallback limit, skip_special_tokens support, SFT re…
oscarqjh Apr 10, 2026
8fb7f22
fix: image truncation retry, interleaved SFT prompt
oscarqjh Apr 13, 2026
2046d20
fix: show image positions inline in prompt log ([img_N] markers)
oscarqjh Apr 13, 2026
bbb8e4b
fix: stop feedback uses successes flag instead of stale geo_dis after…
oscarqjh Apr 16, 2026
a56213f
feat: add maps.zip to dataset zip_files for auto-extraction on download
oscarqjh Apr 16, 2026
5af6400
fix: correct --data-dir help text (default is ~/.cache/easi/datasets,…
oscarqjh Apr 16, 2026
43eb263
fix: revert --data-dir help text (default is ./datasets from runner, …
oscarqjh Apr 16, 2026
722aa5f
refactor: unify default datasets dir — single source of truth in path…
oscarqjh Apr 16, 2026
01cfc3a
feat: added mirror experiment for lh-vln
oscarqjh Apr 20, 2026
a213da8
feat(trajectory): record prompt text per step
oscarqjh Apr 22, 2026
3d1e41e
feat(lhpr_vln/mirror): dump per-step flipped frames via MIRROR_DEBUG_DIR
oscarqjh Apr 22, 2026
e9aca39
fix(lhpr_vln): default skip_special_tokens=false
oscarqjh Apr 22, 2026
861e5a2
feat(lhpr_vln/sft): builder-side enhanced variant
oscarqjh Apr 22, 2026
4f0ad0b
feat(lhpr_vln/sft): bridge-side enhanced variant
oscarqjh Apr 22, 2026
b6eb3c6
fix(lhpr_vln/bridge): remove extra move_forward on reset
oscarqjh Apr 22, 2026
36237ef
feat(lhpr_vln): short experiment task lhpr_vln_val_exp
oscarqjh Apr 22, 2026
2a97532
fix(lhpr_vln): drop step==-1 early return so max_step timeout fires
oscarqjh Apr 22, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
17 changes: 16 additions & 1 deletion .gitignore
Original file line number Diff line number Diff line change
@@ -1 +1,16 @@
/logs
/logs
*.pyc
__pycache__/
*.pyo
*.egg-info
/build
/datasets
/results
/BEHAVIOR-1K
.worktrees/
/docs/plans
yolo11n.pt
.tmp*
/scripts
.sisyphus/
.claude/
Loading