Self-Feeding Chat bot Bug Whilst Interaction

[Dogy06]

There are two issues:

1. Every two messages, the bot replies with: [SelfFeeding]: Oops! I think I messed up. What could I have said instead? no matter what I say. Is this a bug or is this intended?
2. Whenever I try the [DONE] command, the issue below pops up.

I used the pretrained model and trained my own model, both gave same results

Training command: python examples/train_model.py -t self_feeding:all --model-file /tmp/mymodel4 -ltim 5 -vtim 10 -vp 50 -m projects.self_feeding.self_feeding_agent:SelfFeedingAgent -cands batch --eval-candidates inline -histsz 2 --embedding-type fasttext_cc --embedding-size 300 --dict-maxtokens 250000 --num-epochs 500 --optimizer adamax --embeddings-scale false -bs 128 --relu-dropout 0 --attention-dropout 0 --n-heads 2 --n-layers 2 -lr 0.0025 --ffn-size 32 --lr-scheduler invsqrt --warmup-updates 500 --dia-train train_hh131k_hb60k.txt -vmt dia_acc -vmm max

Help would be greatly appreciated. Thanks.

python projects/self_feeding/interactive.py --model-file /tmp/mymodel4 --no-cuda --request-feedback true [ warning: overriding opt['interactive'] to True (previously: False )] [ warning: overriding opt['interactive_task'] to True (previously: None )] [ warning: overriding opt['prev_response_filter'] to True (previously: False )] [ warning: overriding opt['partial_load'] to True (previously: False )] [ warning: overriding opt['eval_candidates'] to fixed (previously: inline )] [ warning: overriding opt['fixed_candidates_path'] to /home/~/ParlAI/data/self_feeding/convai2_cands.txt (previously: None )] [ warning: overriding opt['no_cuda'] to True (previously: False )] [ warning: overriding opt['request_feedback'] to True (previously: False )] [ Setting interactive mode defaults... ] Dictionary: loading dictionary from /tmp/mymodel4.dict [ num words = 25141 ] [SelfFeeding: full interactive mode on.] Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Total parameters: 18,298,093 (17,376,493 trainable) Loading existing model parameters from /tmp/mymodel4 [ Loading fixed candidate set from /home/~/ParlAI/data/self_feeding/convai2_cands.txt ] [ Loading fixed candidate set vectors from /tmp/mymodel4.convai2_cands.vecs ] [ Loading fixed candidate set encodings from /tmp/mymodel4.convai2_cands.encs ] /home/~/.local/lib/python3.7/site-packages/torch/nn/_reduction.py:43: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead. warnings.warn(warning.format(ret)) [creating task(s): parlai.agents.local_human.local_human:LocalHumanAgent] Enter [DONE] if you want to end the episode, [EXIT] to quit. [ optional arguments: ] [ share_word_embeddings: True ] [ Main ParlAI Arguments: ] [ batchsize: 128 ] [ datapath: /home/~/ParlAI/data ] [ datatype: train ] [ download_path: /home/~/ParlAI/downloads ] [ dynamic_batching: None ] [ hide_labels: False ] [ image_mode: raw ] [ init_opt: None ] [ multitask_weights: [1] ] [ numthreads: 1 ] [ show_advanced_args: False ] [ task: self_feeding:all ] [ ParlAI Model Arguments: ] [ dict_class: parlai.core.dict:DictionaryAgent ] [ init_model: None ] [ model: projects.self_feeding.self_feeding_agent:SelfFeedingAgent ] [ model_file: /tmp/mymodel4 ] [ Local Human Arguments: ] [ local_human_candidates_file: None ] [ single_turn: False ] [ ParlAI Image Preprocessing Arguments: ] [ image_cropsize: 224 ] [ image_size: 256 ] [ TorchAgent Arguments: ] [ add_p1_after_newln: False ] [ delimiter: ] [ embedding_projection: random ] [ embedding_type: fasttext_cc ] [ force_fp16_tokens: False ] [ fp16: False ] [ fp16_impl: apex ] [ gpu: -1 ] [ history_add_global_end_token: None ] [ history_size: 2 ] [ interactive_mode: True ] [ label_truncate: None ] [ no_cuda: True ] [ person_tokens: False ] [ rank_candidates: True ] [ split_lines: False ] [ text_truncate: None ] [ truncate: 1024 ] [ use_reply: label ] [ Optimizer Arguments: ] [ adafactor_eps: [1e-30, 0.001] ] [ adam_eps: 1e-08 ] [ betas: [0.9, 0.999] ] [ gradient_clip: 0.1 ] [ learningrate: 0.0025 ] [ momentum: 0 ] [ nesterov: True ] [ nus: [0.7] ] [ optimizer: adamax ] [ weight_decay: None ] [ Learning Rate Scheduler: ] [ invsqrt_lr_decay_gamma: -1 ] [ lr_scheduler: invsqrt ] [ lr_scheduler_decay: 0.5 ] [ lr_scheduler_patience: 3 ] [ max_lr_steps: -1 ] [ update_freq: 1 ] [ warmup_rate: 0.0001 ] [ warmup_updates: 500 ] [ TorchRankerAgent: ] [ candidates: batch ] [ cap_num_predictions: 100 ] [ encode_candidate_vecs: True ] [ encode_candidate_vecs_batchsize: 256 ] [ eval_candidates: fixed ] [ fixed_candidate_vecs: reuse ] [ fixed_candidates_path: /home/~/ParlAI/data/self_feeding/convai2_cands.txt ] [ ignore_bad_candidates: False ] [ inference: max ] [ init_model: None ] [ interactive_candidates: fixed ] [ rank_top_k: -1 ] [ repeat_blocking_heuristic: True ] [ return_cand_scores: False ] [ topk: 5 ] [ train_predict: False ] [ Transformer Arguments: ] [ activation: relu ] [ attention_dropout: 0.0 ] [ data_parallel: False ] [ dropout: 0.0 ] [ embedding_size: 300 ] [ embeddings_scale: False ] [ ffn_size: 32 ] [ learn_embeddings: True ] [ learn_positional_embeddings: False ] [ memory_attention: sqrt ] [ model_parallel: False ] [ n_decoder_layers: -1 ] [ n_encoder_layers: -1 ] [ n_heads: 2 ] [ n_layers: 2 ] [ n_positions: None ] [ n_segments: 0 ] [ normalize_sent_emb: False ] [ output_scaling: 1.0 ] [ reduction_type: mean ] [ relu_dropout: 0.0 ] [ share_encoders: True ] [ use_memories: False ] [ variant: aiayn ] [ wrap_memory_encoder: False ] [ Dictionary Arguments: ] [ bpe_debug: False ] [ dict_endtoken: end ] [ dict_file: /tmp/mymodel4.dict ] [ dict_initpath: None ] [ dict_language: english ] [ dict_lower: False ] [ dict_max_ngram_size: -1 ] [ dict_maxtokens: 250000 ] [ dict_minfreq: 0 ] [ dict_nulltoken: null ] [ dict_starttoken: start ] [ dict_textfields: text,labels ] [ dict_tokenizer: re ] [ dict_unktoken: unk ] [ BPEHelper Arguments: ] [ bpe_add_prefix_space: None ] [ bpe_merge: None ] [ bpe_vocab: None ] [ Self Feeding Model: ] [ sat_head_layers: 1 ] [ share_fee_embeddings: True ] [ share_fee_x_encoder: True ] [ share_fee_y_encoder: True ] [ share_sat_embeddings: False ] [ share_sat_encoder: False ] [ Self-feeding Agent: ] [ add_double_person_tokens: False ] [ dia_weight: 1.0 ] [ display_sat_estimate: False ] [ fee_weight: 1.0 ] [ freeze_base: False ] [ interactive: True ] [ partial_load: True ] [ prev_response_filter: True ] [ prev_response_negatives: False ] [ rating_frequency: 0.01 ] [ rating_gap: 0.05 ] [ rating_threshold: 0.5 ] [ request_feedback: True ] [ request_rating: False ] [ sat_weight: 1.0 ] [ subtasks: ['dialog', 'feedback', 'satisfaction'] ] [ target_class: 0 ] [ Self-feeding Variants: ] [ regex: False ] [ uncertainty_predictor: False ] [ uncertainty_style: gap ] [ uncertainty_threshold: 0.5 ] Enter Your Message: Hello! /home/~/ParlAI/parlai/core/torch_ranker_agent.py:759: UserWarning: [ Executing eval mode with a common set of fixed candidates (n = 127712). ] "(n = {}). ]".format(mode, len(self.fixed_candidates)) [SelfFeeding]: helloooo ! ! Enter Your Message: How are you? [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: You did not mess up [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: Sure, how about Science? [SelfFeeding]: not bad . how about you ? Enter Your Message: hmmm wrong answer [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: You could have said something relating to science [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: a [SelfFeeding]: i like to spend time with my family Enter Your Message: b [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: c [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: d [SelfFeeding]: what is d d ? Enter Your Message: e [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: f [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: g [SelfFeeding]: savage ? do you like death metal ? i know a member of martius . Enter Your Message: [DONE] Traceback (most recent call last): File "projects/self_feeding/interactive.py", line 90, in interactive(parser.parse_args(print_args=False), print_parser=parser) File "projects/self_feeding/interactive.py", line 81, in interactive world.parley() File "/home/~/ParlAI/parlai/core/worlds.py", line 342, in parley acts[1] = agents[1].act() File "/home/~/ParlAI/projects/self_feeding/self_feeding_agent.py", line 324, in act positivity = self.predict_satisfaction(self.observation) File "/home/~/ParlAI/projects/self_feeding/self_feeding_agent.py", line 732, in predict_satisfaction prob = self.model.score_satisfaction(observation['text_vec'].reshape(1, -1)) File "/home/~/ParlAI/projects/self_feeding/modules.py", line 148, in score_satisfaction return torch.sigmoid(self.x_sat_head(self.x_sat_encoder(x_vecs))).squeeze(1) File "/home/~/.local/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547, in call result = self.forward(*input, **kwargs) File "/home/~/ParlAI/parlai/agents/transformer/modules.py", line 522, in forward if positions.max().item() > self.n_positions: RuntimeError: invalid argument 1: cannot perform reduction function max on tensor with no elements because the operation does not have an identity at /pytorch/aten/src/TH/generic/THTensorEvenMoreMath.cpp:189

Another error when using [DONE] command and then typing something: Traceback (most recent call last): File "projects/self_feeding/interactive.py", line 90, in interactive(parser.parse_args(print_args=False), print_parser=parser) File "projects/self_feeding/interactive.py", line 81, in interactive world.parley() File "/home/~/ParlAI/parlai/core/worlds.py", line 341, in parley agents[1].observe(validate(acts[0])) File "/home/~/ParlAI/projects/self_feeding/self_feeding_agent.py", line 372, in observe return super().observe(observation) File "/home/~/ParlAI/parlai/core/torch_agent.py", line 1590, in observe self._validate_observe_invariants() File "/home/~/ParlAI/parlai/core/torch_agent.py", line 1680, in _validate_observe_invariants "Last observe() was episode_done, but we never saw a corresponding " RuntimeError: Last observe() was episode_done, but we never saw a corresponding self_observe to clear the history, probably because you missed an act(). This was changed in #2043. File a GitHub issue if you require assistance.

[Dogy06]

Sorry, Im not sure how some of the error stack trace is crossed out. Sorry for the inconvenience.

[Dogy06]

There are two issues:

Every two messages, the bot replies with: [SelfFeeding]: Oops! I think I messed up. What could I have said instead? no matter what I say. Is this a bug or is this intended? Whenever I try the [DONE] command, the issue below pops up. I used the pretrained model and trained my own model, both gave same results

Training command: python examples/train_model.py -t self_feeding:all --model-file /tmp/mymodel4 -ltim 5 -vtim 10 -vp 50 -m projects.self_feeding.self_feeding_agent:SelfFeedingAgent -cands batch --eval-candidates inline -histsz 2 --embedding-type fasttext_cc --embedding-size 300 --dict-maxtokens 250000 --num-epochs 500 --optimizer adamax --embeddings-scale false -bs 128 --relu-dropout 0 --attention-dropout 0 --n-heads 2 --n-layers 2 -lr 0.0025 --ffn-size 32 --lr-scheduler invsqrt --warmup-updates 500 --dia-train train_hh131k_hb60k.txt -vmt dia_acc -vmm max

Help would be greatly appreciated. Thanks.

python projects/self_feeding/interactive.py --model-file /tmp/mymodel4 --no-cuda --request-feedback true [ warning: overriding opt['interactive'] to True (previously: False )] [ warning: overriding opt['interactive_task'] to True (previously: None )] [ warning: overriding opt['prev_response_filter'] to True (previously: False )] [ warning: overriding opt['partial_load'] to True (previously: False )] [ warning: overriding opt['eval_candidates'] to fixed (previously: inline )] [ warning: overriding opt['fixed_candidates_path'] to /home//ParlAI/data/self_feeding/convai2_cands.txt (previously: None )] [ warning: overriding opt['no_cuda'] to True (previously: False )] [ warning: overriding opt['request_feedback'] to True (previously: False )] [ Setting interactive mode defaults... ] Dictionary: loading dictionary from /tmp/mymodel4.dict [ num words = 25141 ] [SelfFeeding: full interactive mode on.] Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Initialized embeddings for 23672 tokens (94.2%) from fasttext_cc. Total parameters: 18,298,093 (17,376,493 trainable) Loading existing model parameters from /tmp/mymodel4 [ Loading fixed candidate set from /home//ParlAI/data/self_feeding/convai2_cands.txt ] [ Loading fixed candidate set vectors from /tmp/mymodel4.convai2_cands.vecs ] [ Loading fixed candidate set encodings from /tmp/mymodel4.convai2_cands.encs ] /home/~/.local/lib/python3.7/site-packages/torch/nn/_reduction.py:43: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead. warnings.warn(warning.format(ret)) [creating task(s): parlai.agents.local_human.local_human:LocalHumanAgent] Enter [DONE] if you want to end the episode, [EXIT] to quit. [ optional arguments: ] [ share_word_embeddings: True ] [ Main ParlAI Arguments: ] [ batchsize: 128 ] [ datapath: /home//ParlAI/data ] [ datatype: train ] [ download_path: /home//ParlAI/downloads ] [ dynamic_batching: None ] [ hide_labels: False ] [ image_mode: raw ] [ init_opt: None ] [ multitask_weights: [1] ] [ numthreads: 1 ] [ show_advanced_args: False ] [ task: self_feeding:all ] [ ParlAI Model Arguments: ] [ dict_class: parlai.core.dict:DictionaryAgent ] [ init_model: None ] [ model: projects.self_feeding.self_feeding_agent:SelfFeedingAgent ] [ model_file: /tmp/mymodel4 ] [ Local Human Arguments: ] [ local_human_candidates_file: None ] [ single_turn: False ] [ ParlAI Image Preprocessing Arguments: ] [ image_cropsize: 224 ] [ image_size: 256 ] [ TorchAgent Arguments: ] [ add_p1_after_newln: False ] [ delimiter: ] [ embedding_projection: random ] [ embedding_type: fasttext_cc ] [ force_fp16_tokens: False ] [ fp16: False ] [ fp16_impl: apex ] [ gpu: -1 ] [ history_add_global_end_token: None ] [ history_size: 2 ] [ interactive_mode: True ] [ label_truncate: None ] [ no_cuda: True ] [ person_tokens: False ] [ rank_candidates: True ] [ split_lines: False ] [ text_truncate: None ] [ truncate: 1024 ] [ use_reply: label ] [ Optimizer Arguments: ] [ adafactor_eps: [1e-30, 0.001] ] [ adam_eps: 1e-08 ] [ betas: [0.9, 0.999] ] [ gradient_clip: 0.1 ] [ learningrate: 0.0025 ] [ momentum: 0 ] [ nesterov: True ] [ nus: [0.7] ] [ optimizer: adamax ] [ weight_decay: None ] [ Learning Rate Scheduler: ] [ invsqrt_lr_decay_gamma: -1 ] [ lr_scheduler: invsqrt ] [ lr_scheduler_decay: 0.5 ] [ lr_scheduler_patience: 3 ] [ max_lr_steps: -1 ] [ update_freq: 1 ] [ warmup_rate: 0.0001 ] [ warmup_updates: 500 ] [ TorchRankerAgent: ] [ candidates: batch ] [ cap_num_predictions: 100 ] [ encode_candidate_vecs: True ] [ encode_candidate_vecs_batchsize: 256 ] [ eval_candidates: fixed ] [ fixed_candidate_vecs: reuse ] [ fixed_candidates_path: /home//ParlAI/data/self_feeding/convai2_cands.txt ] [ ignore_bad_candidates: False ] [ inference: max ] [ init_model: None ] [ interactive_candidates: fixed ] [ rank_top_k: -1 ] [ repeat_blocking_heuristic: True ] [ return_cand_scores: False ] [ topk: 5 ] [ train_predict: False ] [ Transformer Arguments: ] [ activation: relu ] [ attention_dropout: 0.0 ] [ data_parallel: False ] [ dropout: 0.0 ] [ embedding_size: 300 ] [ embeddings_scale: False ] [ ffn_size: 32 ] [ learn_embeddings: True ] [ learn_positional_embeddings: False ] [ memory_attention: sqrt ] [ model_parallel: False ] [ n_decoder_layers: -1 ] [ n_encoder_layers: -1 ] [ n_heads: 2 ] [ n_layers: 2 ] [ n_positions: None ] [ n_segments: 0 ] [ normalize_sent_emb: False ] [ output_scaling: 1.0 ] [ reduction_type: mean ] [ relu_dropout: 0.0 ] [ share_encoders: True ] [ use_memories: False ] [ variant: aiayn ] [ wrap_memory_encoder: False ] [ Dictionary Arguments: ] [ bpe_debug: False ] [ dict_endtoken: end ] [ dict_file: /tmp/mymodel4.dict ] [ dict_initpath: None ] [ dict_language: english ] [ dict_lower: False ] [ dict_max_ngram_size: -1 ] [ dict_maxtokens: 250000 ] [ dict_minfreq: 0 ] [ dict_nulltoken: null ] [ dict_starttoken: start ] [ dict_textfields: text,labels ] [ dict_tokenizer: re ] [ dict_unktoken: unk ] [ BPEHelper Arguments: ] [ bpe_add_prefix_space: None ] [ bpe_merge: None ] [ bpe_vocab: None ] [ Self Feeding Model: ] [ sat_head_layers: 1 ] [ share_fee_embeddings: True ] [ share_fee_x_encoder: True ] [ share_fee_y_encoder: True ] [ share_sat_embeddings: False ] [ share_sat_encoder: False ] [ Self-feeding Agent: ] [ add_double_person_tokens: False ] [ dia_weight: 1.0 ] [ display_sat_estimate: False ] [ fee_weight: 1.0 ] [ freeze_base: False ] [ interactive: True ] [ partial_load: True ] [ prev_response_filter: True ] [ prev_response_negatives: False ] [ rating_frequency: 0.01 ] [ rating_gap: 0.05 ] [ rating_threshold: 0.5 ] [ request_feedback: True ] [ request_rating: False ] [ sat_weight: 1.0 ] [ subtasks: ['dialog', 'feedback', 'satisfaction'] ] [ target_class: 0 ] [ Self-feeding Variants: ] [ regex: False ] [ uncertainty_predictor: False ] [ uncertainty_style: gap ] [ uncertainty_threshold: 0.5 ] Enter Your Message: Hello! /home//ParlAI/parlai/core/torch_ranker_agent.py:759: UserWarning: [ Executing eval mode with a common set of fixed candidates (n = 127712). ] "(n = {}). ]".format(mode, len(self.fixed_candidates)) [SelfFeeding]: helloooo ! ! Enter Your Message: How are you? [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: You did not mess up [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: Sure, how about Science? [SelfFeeding]: not bad . how about you ? Enter Your Message: hmmm wrong answer [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: You could have said something relating to science [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: a [SelfFeeding]: i like to spend time with my family Enter Your Message: b [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: c [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: d [SelfFeeding]: what is d d ? Enter Your Message: e [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: f [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: g [SelfFeeding]: savage ? do you like death metal ? i know a member of martius . Enter Your Message: [DONE] Traceback (most recent call last): File "projects/self_feeding/interactive.py", line 90, in interactive(parser.parse_args(print_args=False), print_parser=parser) File "projects/self_feeding/interactive.py", line 81, in interactive world.parley() File "/home//ParlAI/parlai/core/worlds.py", line 342, in parley acts[1] = agents[1].act() File "/home//ParlAI/projects/self_feeding/self_feeding_agent.py", line 324, in act positivity = self.predict_satisfaction(self.observation) File "/home//ParlAI/projects/self_feeding/self_feeding_agent.py", line 732, in predict_satisfaction prob = self.model.score_satisfaction(observation['text_vec'].reshape(1, -1)) File "/home//ParlAI/projects/self_feeding/modules.py", line 148, in score_satisfaction return torch.sigmoid(self.x_sat_head(self.x_sat_encoder(x_vecs))).squeeze(1) File "/home//.local/lib/python3.7/site-packages/torch/nn/modules/module.py", line 547, in call result = self.forward(*input, **kwargs) File "/home//ParlAI/parlai/agents/transformer/modules.py", line 522, in forward if positions.max().item() > self.n_positions: RuntimeError: invalid argument 1: cannot perform reduction function max on tensor with no elements because the operation does not have an identity at /pytorch/aten/src/TH/generic/THTensorEvenMoreMath.cpp:189

Another error when using [DONE] command and then typing something: Traceback (most recent call last): File "projects/self_feeding/interactive.py", line 90, in interactive(parser.parse_args(print_args=False), print_parser=parser) File "projects/self_feeding/interactive.py", line 81, in interactive world.parley() File "/home//ParlAI/parlai/core/worlds.py", line 341, in parley agents[1].observe(validate(acts[0])) File "/home//ParlAI/projects/self_feeding/self_feeding_agent.py", line 372, in observe return super().observe(observation) File "/home//ParlAI/parlai/core/torch_agent.py", line 1590, in observe self._validate_observe_invariants() File "/home//ParlAI/parlai/core/torch_agent.py", line 1680, in _validate_observe_invariants "Last observe() was episode_done, but we never saw a corresponding " RuntimeError: Last observe() was episode_done, but we never saw a corresponding self_observe to clear the history, probably because you missed an act(). This was changed in #2043. File a GitHub issue if you require assistance.

[stephenroller]

Thanks for reporting. The second one looks like there was a batch size of 0 maybe? The first one I'm not so sure about. We'd have to figure out why there was the observe call but no self observe. Perhaps the agent's act wasn't upgraded...

[Dogy06]

Oh ok. Thanks for your help. Regarding the first problem, should I try an earlier commit of this repository.

Thanks again.

[stephenroller]

Did you happen to have luck with an earlier commit? If so, do you know which? That would help me fix the bug in master.

[Dogy06]

Ah ok. I will try an earlier commit and come back to you.

[Dogy06]

Even after trying commit as early as #2124, I still couldn't get it to work. It may be a problem with my parameters. Command I used:python projects/self_feeding/interactive.py --model-file zoo:self_feeding/hh131k_hb60k_fb60k_st1k/model --no-cuda --request-feedback true Still getting this: Enter Your Message: Hello [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: [SelfFeeding]: good morning what is the haps ? Enter Your Message: [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message: [SelfFeeding]: Thanks! I'll try to remember that. Can you pick a new topic for us to talk about now? Enter Your Message: [SelfFeeding]: what is up stranger ? what are you doing ? Enter Your Message: hi [SelfFeeding]: Oops! I think I messed up. What could I have said instead? Enter Your Message:

parameters: [ optional arguments: ] [ share_word_embeddings: True ] [ Main ParlAI Arguments: ] [ batchsize: 128 ] [ datapath: /home/weichen/ParlAI/data ] [ datatype: train ] [ download_path: /dfs/scratch1/bradenjh/ParlAI/downloads ] [ hide_labels: False ] [ image_mode: raw ] [ init_opt: None ] [ multitask_weights: [1] ] [ numthreads: 1 ] [ show_advanced_args: False ] [ task: self_feeding:all:train ] [ ParlAI Model Arguments: ] [ dict_class: parlai.core.dict:DictionaryAgent ] [ init_model: None ] [ model: projects.self_feeding.self_feeding_agent:SelfFeedingAgent ] [ model_file: /home/weichen/ParlAI/data/models/self_feeding/hh131k_hb60k_fb60k_st1k/model ] [ Local Human Arguments: ] [ local_human_candidates_file: None ] [ single_turn: False ] [ ParlAI Image Preprocessing Arguments: ] [ image_cropsize: 224 ] [ image_size: 256 ] [ TorchAgent Arguments: ] [ add_p1_after_newln: False ] [ delimiter: ] [ embedding_projection: random ] [ embedding_type: fasttext_cc ] [ fp16: False ] [ gpu: -1 ] [ history_size: 2 ] [ interactive_mode: True ] [ label_truncate: None ] [ no_cuda: True ] [ person_tokens: False ] [ rank_candidates: True ] [ split_lines: False ] [ text_truncate: None ] [ truncate: 1024 ] [ use_reply: label ] [ Optimizer Arguments: ] [ adam_eps: 1e-08 ] [ betas: [0.9, 0.999] ] [ gradient_clip: 0.1 ] [ learningrate: 0.0025 ] [ momentum: 0 ] [ nesterov: True ] [ nus: [0.7] ] [ optimizer: adamax ] [ weight_decay: None ] [ Learning Rate Scheduler: ] [ lr_scheduler: invsqrt ] [ lr_scheduler_decay: 0.5 ] [ lr_scheduler_patience: 3 ] [ update_freq: -1 ] [ warmup_rate: 0.0001 ] [ warmup_updates: 500 ] [ TorchRankerAgent: ] [ candidates: batch ] [ cap_num_predictions: 100 ] [ encode_candidate_vecs: True ] [ encode_candidate_vecs_batchsize: 256 ] [ eval_candidates: fixed ] [ fixed_candidate_vecs: reuse ] [ fixed_candidates_path: /home/weichen/ParlAI/data/self_feeding/convai2_cands.txt ] [ ignore_bad_candidates: False ] [ inference: max ] [ init_model: None ] [ rank_top_k: -1 ] [ repeat_blocking_heuristic: True ] [ topk: 5 ] [ train_predict: False ] [ Transformer Arguments: ] [ activation: relu ] [ attention_dropout: 0.0 ] [ data_parallel: False ] [ dropout: 0.0 ] [ embedding_size: 300 ] [ embeddings_scale: False ] [ ffn_size: 32 ] [ learn_embeddings: True ] [ learn_positional_embeddings: False ] [ memory_attention: sqrt ] [ n_heads: 2 ] [ n_layers: 2 ] [ n_positions: None ] [ n_segments: 0 ] [ normalize_sent_emb: False ] [ output_scaling: 1.0 ] [ reduction_type: mean ] [ relu_dropout: 0.0 ] [ share_encoders: True ] [ use_memories: False ] [ variant: aiayn ] [ wrap_memory_encoder: False ] [ Dictionary Arguments: ] [ bpe_debug: False ] [ dict_endtoken: end ] [ dict_file: /home/weichen/ParlAI/data/models/self_feeding/hh131k_hb60k_fb60k_st1k/model.dict ] [ dict_initpath: None ] [ dict_language: english ] [ dict_lower: False ] [ dict_max_ngram_size: -1 ] [ dict_maxtokens: 250000 ] [ dict_minfreq: 0 ] [ dict_nulltoken: null ] [ dict_starttoken: start ] [ dict_textfields: text,labels ] [ dict_tokenizer: re ] [ dict_unktoken: unk ] [ Self Feeding Model: ] [ sat_head_layers: 1 ] [ share_fee_embeddings: True ] [ share_fee_x_encoder: True ] [ share_fee_y_encoder: True ] [ share_sat_embeddings: False ] [ share_sat_encoder: False ] [ Self-feeding Agent: ] [ add_double_person_tokens: True ] [ dia_weight: 1.0 ] [ display_sat_estimate: False ] [ fee_weight: 1.0 ] [ freeze_base: False ] [ interactive: True ] [ partial_load: True ] [ prev_response_filter: True ] [ prev_response_negatives: False ] [ rating_frequency: 0.01 ] [ rating_gap: 0.05 ] [ rating_threshold: 0.5 ] [ request_feedback: True ] [ request_rating: False ] [ sat_weight: 1.0 ] [ subtasks: ['dialog', 'satisfaction'] ] [ target_class: 0 ] [ Self-feeding Variants: ] [ regex: False ] [ uncertainty_predictor: False ] [ uncertainty_style: gap ] [ uncertainty_threshold: 0.5 ]

[github-actions[bot]]

This issue has not had activity in 30 days. Marking as stale.

[klshuster]

closing, as this project is no longer maintained