# C5W1 A1 (Ex8) lstm\_backward, dc\_next missing? **URL:** \ **Category:** Sequence Models\ **Tags:** coursera-platform\ **Created:** [November 12, 2021, 7:31am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321 "2021-11-12T07:31:22Z")\ **Posts on this page:** 14\ **Page:** 1
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 12, 2021, 7:31am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/1 "2021-11-12T07:31:22Z")
I have a trouble merging lstm\_cell\_backward in lstm\_backward because the former requires dc\_next but the latter exercise(lstm\_backward) does not provide it. Maybe I have missed something. Please advise me. Thanks, Youngha ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 17, 2021, 6:24pm UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/2 "2021-11-17T18:24:02Z")
How did you solve this in rnn\_backward? You see that da\_prev is da\_next for the previous timestep. The same is true for dc\_prev and dc\_next. ![Screen Shot 2021-11-17 at 19.21.31](https://global.discourse-cdn.com/dlai/original/2X/d/d5c1bf41cdc585f5f95fe65b51407c24c7109990.jpeg) ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 18, 2021, 8:43am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/4 "2021-11-18T08:43:17Z")
![Screen Shot 2021-11-18 at 09.30.08](https://global.discourse-cdn.com/dlai/original/2X/3/3916f75aee2038a866c46c914b68ba1186993537.png) You also need to take into account da[:,:,t] when calling your backward cell functions. ---
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 18, 2021, 9:49am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/5 "2021-11-18T09:49:50Z")
I got that part but I have an broadcast error calling lstm\_cell\_backward apparently from the argument da[:,:,t]+da\_prevt. I couldn’t figure out the reason yet. ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 18, 2021, 10:09am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/6 "2021-11-18T10:09:27Z")
Your call looks correct, so maybe you have a problem where you update da\_prevt, which gives incompatible dimensions. ---
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 18, 2021, 10:36am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/7 "2021-11-18T10:36:18Z")
you’re right but I don’t understand because Wf^T \* dft gave the dimension as [8, 10]. That makes sense from 8 = 5 + 3, where n\_a = 5, and n\_x = 3. Wf was extracted by parameters[“Wf”]. Maybe eq. 19 was not correct? ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 18, 2021, 10:40am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/8 "2021-11-18T10:40:21Z")
Can I have a look at your implementation? ---
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 18, 2021, 10:48am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/9 "2021-11-18T10:48:19Z")
_{moderator edit - solution code removed}_ ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 18, 2021, 11:01am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/10 "2021-11-18T11:01:06Z")
Can you also supply lstm\_backward(da, caches) ---
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 18, 2021, 12:01pm UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/11 "2021-11-18T12:01:33Z")
_{moderator edit - solution code removed}_ ---
**Author:** ![jonaslalin](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/jonaslalin/32/63_2.png) [@jonaslalin](https://community.deeplearning.ai/u/jonaslalin)\ **Post date:** [November 18, 2021, 12:13pm UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/12 "2021-11-18T12:13:13Z")
> [@Youngha\_Hwang](#): > > ```auto > da_prev = np.dot(parameters['Wf'].T, dft) + np.dot(parameters['Wi'].T, dit) + np.dot(parameters['Wc'].T, dcct) + np.dot(parameters['Wo'].T, dot) > > ``` is not correct. ![Screen Shot 2021-11-18 at 13.12.20](https://global.discourse-cdn.com/dlai/original/2X/3/3e0b9f899e477cc89516ff5b5dbbf65782fc2f43.png) ---
**Author:** ![Youngha\_Hwang](https://avatars.discourse-cdn.com/v4/letter/y/65b543/32.png) [@Youngha\_Hwang](https://community.deeplearning.ai/u/Youngha_Hwang)\ **Post date:** [November 18, 2021, 12:21pm UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/13 "2021-11-18T12:21:29Z")
Oh my bad. Thank you so much. It was the last one assignment that I was working on. ---
**Author:** ![zac\_builta](https://avatars.discourse-cdn.com/v4/letter/z/82dd89/32.png) [@zac\_builta](https://community.deeplearning.ai/u/zac_builta)\ **Post date:** [May 4, 2023, 2:51am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/14 "2023-05-04T02:51:00Z")
Doesn’t that instruction seem unclear or wrong though? like what is W[:,: na] is it missing a comma or is there something I’m unaware of? ---
**Author:** ![saifkhanengr](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/saifkhanengr/32/152858_2.png) [@saifkhanengr](https://community.deeplearning.ai/u/saifkhanengr)\ **Post date:** [May 4, 2023, 5:00am UTC](https://community.deeplearning.ai/t/c5w1-a1-ex8-lstm-backward-dc-next-missing/63321/15 "2023-05-04T05:00:07Z")
I’ve replied to your original post. Kindly check it [here](https://community.deeplearning.ai/t/dls-course-5-week-1-assignment-1-lstm-backward-pass-equations/324270/2). Best, Saif.